Presumed cancer‐associated retinopathy (CAR) mimicking Sudden Acquired Retinal Degeneration Syndrome (SARDS) in canines

Abstract Objective To describe functional and structural features of presumed cancer‐associated retinopathy (CAR) mimicking sudden acquired retinal degeneration syndrome (SARDS) in dogs and describe treatment outcomes. Animals Subjects were 17 dogs from 8 eight US states and Canada diagnosed with SARDS or immune‐mediated retinitis (IMR) by 12 ophthalmologists. Nine eyes from seven deceased patients were used for microarray (MA), histology, or immunohistochemical (IHC) analysis. Procedures Dogs underwent complete ophthalmic examination, including retinal photography, optical coherence tomography (OCT), chromatic pupil light reflex testing (cPLR), and electroretinography (ERG), in addition to complete systemic examination. Histology, microarray, and IHC analysis were performed in CAR retinas to evaluate histological and molecular changes in retinal tissue. Results None of the patients evaluated satisfied previously established criteria for diagnosis of SARDS (flat ERG+ no red – good blue PLR), and all were diagnosed with IMR. All patients were diagnosed with a cancer: meningioma (24%), sarcoma (18%), pituitary tumor (12%), and squamous cell carcinoma (12%), other (34%). Median survival time was 6 months from diagnosis (range 1‐36 months). Most frequent systemic abnormalities were as follows: proteinuria (78%); elevated liver enzymes (47%); and metabolic changes (PU/PD, polyphagia – 24%). Immunosuppressive therapy resulted in the reversal of blindness in 44% of treated patients, with 61% of all treated patients recovering and/or maintaining vision. Median time for preservation of vision was 5 months (range 1‐35 months). Conclusions Observed changes are highly suggestive of immune‐mediated damage in IMR‐CAR eyes. A relatively high percentage of patients with CAR responded positively to immunosuppressive therapy.


| INTRODUCTION
Sudden acquired retinal degeneration syndrome (SARDS) is recognized as one of the most frequent irreversible causes of blindness in the canine population. It is characterized by sudden-onset blindness, completely extinguished retinal electrical responses, and the properties of abnormal chromatic pupil light reflex (cPLR) (no red PLR -good blue PLR). [1][2][3] It has been theorized that SARDs is an autoimmune disease similar to autoimmune retinopathies in humans (AIR), with a strong retinal autoantibody component. [4][5][6][7][8] Autoimmune retinopathies (AIR) are rare but devastating autoimmune diseases in humans, characterized by the sudden onset of severe visual deficits (or complete blindness), extinguished retinal electrical activity, relatively normal fundus appearance, and presence of serum retinal autoantibodies. [9][10][11] AIRs can develop as a form of paraneoplastic syndrome (ie, CAR: cancer-associated retinopathy; MAR: melanoma-associated retinopathy), or in the absence of cancer (nonparaneoplastic autoimmune retinopathies: npAIR). [12][13][14][15] Patients suffering from npAIR can share features of CAR and MAR; however, it is believed that this retinal disease does not develop as a direct result of a cancer. 13 Autoantibodies (autoAbs) against retinal antigens are considered to be the primary cause of pathology in CAR and npAIR. [15][16][17][18] Anti-retinal autoAbs have also been implicated in the pathology of macular degeneration, 17,[19][20][21] retinitis pigmentosa, [20][21][22] and glaucoma. 23,24 In these diseases, however, it is not clear whether the antibody production precedes the retinal disease, or the immune reactivity is a consequence of the retinal degenerative process and exposure of degenerating retinal elements to the immune system. 25 The lack of studies addressing detailed histology and molecular analysis of retinal tissue from CAR human patients further contributes to the relatively poor understanding of mechanisms associated with functional and structural retinal damage. [26][27][28] While there is an extensive body of literature describing the functional and morphological properties of CAR and MAR in humans, 14,27,[29][30][31] a description of CAR is lacking in the veterinary ophthalmology literature. A single review manuscript describes the features of possible immune-mediated retinitis (IMR)-CAR in canine patients. 2 The principal purpose of this study was to describe the clinical, morphological, and molecular properties of presumed CAR in canine patients. Furthermore, due to the extreme similarity between the clinical presentation of SARDS and CAR, the second goal was to describe potential diagnostic parameters that could be used to differentiate between the two conditions. Finally, we evaluate the success of different medical treatment strategies for canine CAR, with a particular focus on a completely novel therapy-the intravitreal application of intravenous immunoglobulin (IVIg).

| MATERIALS AND METHODS
All studies were conducted in accordance with the ARVO Statement for Use of Animals in Ophthalmic and Vision Research. Procedures were approved by the Iowa State University Institutional Committee on Animal Care with protocol numbers: 2-07-6307-K and 5-07-6362-K, while clients pursuing treatment in private hospitals provided a signed consent form for diagnostic and treatment options, including intravitreal IVIg treatment, and possible risks and complications associated with intravitreal injections. A total of 17 canine clinical patients were evaluated for SARDS in the period between September 2006 and April 2017 (Tables 1, 2 and 3). Patients were from eight US states and Canada and were diagnosed with SARDS or immune-mediated retinitis (IMR) by eight different ophthalmologists. Sixteen patients were diagnosed with a malignant tumor prior to or at the time of development of visual problems, while 1 was diagnosed with malignant tumor 24 months after initial development of blindness.
Nine eyes were collected for histology, microarray (MA), or immunohistochemistry (IHC) analysis from seven IMR-CAR patients (Table 1, patient numbers 1, 2, 3, 4, 5, 6, and 7) euthanized within 2 months to 5 years of IMR-CAR diagnosis. One half of the right and one half of the left eye were collected from 2 IMR-CAR patients for microarray analysis (patients number 1 and 2), while remaining halves from these patients were used for histology or IHC analysis. An additional 10 control eyes from healthy control dogs (Beagle, SF, 6 y old) without evidence of ocular abnormalities were used for microarray and immunohistochemistry analysis. Serum samples from 10 glaucomatous Basset Hounds from the Iowa State University colony and 7 healthy Beagles were used for the serum retinal autoantibody detection.

| Diagnosis of the presumed CAR phenotype
Previously published criteria for diagnosis of SARDS were used to characterize all patients: completely extinguished electroretinography (ERG) response (flat ERG), in combination with no pupil light reflex after red light illumination (NR) and good pupil light reflex after blue light illumination (GB). 2 All CAR patients and healthy control dogs received a complete ophthalmic examination: slit-lamp biomicroscopy, indirect ophthalmoscopy, tear production and intraocular pressure evaluation, and basic neuro-ophthalmology evaluation (palpebral and corneal reflex, ocular motility

| Systemic organ function evaluation
All patients also underwent complete cell blood count and serum chemistry, urine analysis (10 of 17, all samples collected via cystocentesis), systolic blood pressure evaluation (10 of 17, systolic blood pressure was evaluated using an ultrasonic Doppler flow detector; Model 811-L, Parks Medical Electronics Inc, Las Vegas, NV, USA), and thoracic and abdominal radiographs (17 of 17). In 7 of 17 patients, brain computerized tomography (CT) or magnetic resonance imaging (MRI) was performed. Abdominal and thoracic radiographs, and brain CT images were reviewed by board-certified radiologists and residents at the institutions where imaging was performed. Only head CT/MRI imaging was pursued (no thorax or abdomen CT imaging was conducted).  Table 1, patient no. 13); (C) completely extinguished low light rod response, reduced combined rod-cone response, with some reduction in cone function (Miniature Pinscher, CM, 7 y old, Table 1, patient no. 6); (D) completely extinguished rod function, with minimal rod-cone response, and bizarre cone responses (increased implicit time for the full-field cone response and absent flicker response) in IMR-CAR canine patient with mammary adenocarcinoma (Springer Spaniel OS, F, 9 y old, patient,

| Functional and structural retinal evaluation in vivo
The vision was evaluated by observing following parameters in dim and bright light conditions: menace, cotton ball tracking, and visual navigation in the maze test. The presence of one or multiple positive visual responses was marked as presence of the vision. Pupil light reflex (PLR), chromatic pupil light reflex (cPLR), fundus photography, and electroretinography (ERG) were performed in clinical patients and experimental dogs as described in previous studies. 2,[32][33][34] Optical coherence tomography (OCT) analysis in canine patients was performed as previously described. 32

| General anesthesia protocol for ERG and OCT testing
Dogs were anesthetized for ERG and OCT testing as previously reported. 5,32,33

| Electroretinography (ERG)
Electroretinography (full-field and pattern routines) was used to evaluate retinal function in IMR-CAR dogs (n = 16). A Roland Consult ERG system (Brandenburg, Germany) was used for pattern ERG routines and the International Society for Clinical Electrophysiology of Vision (ISCEV) ERG routines under general anesthesia, while retinographics ERG system (Norwalk, CT) was used for ERG testing in nonanesthetized and nonsedated patients. Both ERG systems were used to deliver light stimuli and collect signals from the lens electrode for full-field ERG recording routines, as previously reported. 2,5,32

| Multifocal ERG recording protocol
The mfERG routine was recorded after a 15-minute period of dark adaptation using the VERIS Science system (Edi Inc, Redwood City, CA, USA) with fundus camera for projection of the stimulus pattern in one clinical patient. A pediatric bipolar Burian-Allen electrode (Hansen Ophthalmic Development Laboratory, Iowa City, IA) was used to record the mfERG routine. Since a fundus camera with projecting stimuli was used for mfERG recordings, the electrical activity was recorded from only one eye at a time. The reference electrode was positioned in the forehead region between the two eyes, while the ground electrode was placed on the back of the head (occipital region). Both electrodes were inserted subcutaneously.
The first-order response of mfERG was analyzed in CAR patients and control dogs. The stimulus routine consisted of 37 black and white hexagons with temporal settings: frequency = 75 Hz, M sequence = 15, Frames = 1, number of segments = 16, pre-exposure time = 1 s, samples per frame = 16. Low cut amplifier frequency was set at 3 Hz, while high cut amplifier frequency was set at 300 Hz.

| Fundus photography
Fundus photography was performed using the RetCam Fundus Camera system (Massie Research Laboratories, Pleasanton, CA) as previously reported. 5

| Optical coherence tomography (OCT)
Optical coherence tomography and analysis of different retinal layer thickness were performed using Heidelberg Spectralis SD-OCT unit as previously reported. 5,33 The following scans were obtained: horizontal volume scan through area centralis (within the visual streak; located dorsal-temporally from the optic nerve head) in the superior-temporal (tapetal) retina, with a corresponding volume scan in the ventrotemporal (nontapetal) retina. Additional horizontal volume scans were performed based on the funduscopic evidence of possible retinal lesions with a focus on hyperpigmented, hyper-reflective, and potentially exudative lesions. The goal of scans was to detect possible retinal lesions that were not evident on a regular ophthalmoscopy examination, as previously reported for SARDS dogs. 5

| Histological and immunohistochemical (IHC) analysis
Six eyes removed from four euthanized IMR-CAR patients, and six additional eyes from healthy beagles with previously performed ophthalmic examination to rule out presence of the ocular disease were fixed and prepared for histology analysis as previously described. 5,35 Histological and immunohistochemical analysis on six IMR-CAR eyes from three patients (collected 1-35 months after IMR-CAR diagnosis) and six untreated healthy beagle eyes were performed as previously reported using anti-CD3 (T-lymphocyte marker); anti-CD79 (B-lymphocyte marker); anti-CD11b (microglia marker); and a cocktail of IgG, IgM, and IgA for detection of immunoglobulin-producing plasma cells as previously described. 5,35

| Detection of serum and intravitreal retinal autoantibodies-Western blot analysis
Initial screening of dog IMR-CAR sera (n = 5), IMR-CAR vitreal sample (n = 1), control healthy dog canine sera (n = 7), and banked frozen primary glaucoma dog sera (n = 10) was performed using dog retinal proteins that were extracted from retinas as previously described. 36

| Microarray (MA) analysis
Microarray analysis on canine retinal tissue was performed as previously reported. 5 Briefly, immediately after enucleation eyes were preserved in RNAlater (Thermo Fisher). Retinas were then dissected, and RNA was extracted and hybridized to Canine Genome 2.0 arrays (Thermo Fisher) following the manufacturer's instructions. Data were obtained from two eyes of an untreated IMR-CAR patient, two of a IMR-CAR patient treated with intraocular IVIG who responded to therapy, three from three patients with SARDS (previously published SARDS microarray dataset 5 was used for comparison with IMR-CAR microarray data), and two from two healthy control dogs without evidence of ocular abnormalities.

| Statistical analysis
Obtained microarray data were normalized using the robust multi-array average (RMA) module in R (version 3.6). 37 Probesets that yielded low expression values in all samples were removed from the dataset. Hierarchical clustering and PCA (principal component analysis) were carried out using ClustVis software (version 2.0). 38 Functional assignments were generated using DAVID (database for annotation, visualization, and integrated discovery, v6.8) 39 and STRING. 40 Calculation of mean, standard deviation, and medial value and paired t test was performed with commercial software (Prism, version 5.0; GraphPad, San Diego, CA). Value of P < .05 was considered statistically significant.

| Patient population
Subjects were 17 dogs from eight US states and Canada diagnosed with SARDS or IMR based on history and results of ophthalmology and general examination by eight ophthalmologists (Tables 1 and 2).
Sixteen patients were diagnosed with a malignant tumor prior to or at the time of visual problem onset, while one was diagnosed with a malignant tumor 24 months after initial development of blindness (Table 3). The most frequent tumors were as follows: meningioma (n = 4), sarcoma (n = 3), pituitary tumor (n = 2), and squamous cell carcinoma (n = 2). Median survival time was 6 months from diagnosis (range 1-36 months). All patients were deceased by May 2017 (Table 3).

| Functional analysis
A total of 56% (9 of 16) of patients had completely extinguished retinal electrical responses (flat ERG), 18.7% (3 of 16) had normal full-field responses but decreased or absent flicker ERG responses, 18.7% (3 of 16) had decreased full-field responses and flicker ERG responses, while one had near-normal full-field scotopic and photopic responses, but absent rod and pattern ERG responses, and suppressed mfERG responses (Table 1, Figure 1). Chromatic pupil light reflex testing revealed completely absent cPLR after red light illumination and normal cPLR after blue light illumination in 23.5% of patients (4 of 17; Table 1, Figure 2). Two patients had completely normal cPLR responses (both had completely extinguished ERG responses), one had completely absent cPLR response after red and blue light illumination, while the rest had different combination of cPLR deficits in one or both eyes, with no PLR after red light illumination and decreased PLR response after blue light illumination being the most common type of deficit (total of seven eyes). None of these patients had the classic combination of SARDS responses: flat ERG combined with no red-good blue cPLR (Table 1).

| Fundoscopic findings
Fundoscopic evaluation revealed presence of various changes. The most frequently observed fundus change was vascular attenuation (VA) resulting in decreased diameter of blood vessels, and sporadic loss of tertiary retinal vein branches, which was observed in 41% of patients (7 of 17; Table 1, Figure 3). In addition, pale optic nerve head appearance (pONH) was observed in 23.5% of patients (4 of 17; Table 1, Figure 3). Normal fundus was observed in 23.5% of patients (4 of 17), while an additional two had one eye | 135 with normal fundus appearance. Altered tapetal reflectivity, perivascular hyper-reflective lesions, chorioretinal scars, hyperpigmented spots, and retinal edema could be observed in 23.5% of patients (4 of 17). One patient had bilateral optic nerve head papilledema.

| Optical coherence tomography (OCT) and retinal histology analysis
Optical coherence tomography analysis was performed in five patients and comparison to fundus images revealed that zones of hyper-reflective appearance were associated with focal retinal structural photoreceptor loss (Figures 4 and 5). Diffuse loss of the inner segment-outer segment (IS-OS) photoreceptor organization was evident (Figures 4 and 5). Zones of perivascular exudative lesions (PEL) were associated with exudative retinal detachments, retinal edematous, and cystic changes or retinal perivascular thinning (Figures 4 and 5). Histology analysis revealed the presence of photoreceptor loss, which had a focal perivascular nature in some cases, activation of retinal pigment epithelium, and active macrophage appearance ( Figure 5).

| Systemic organ changes
Analysis of systemic changes (increased appetite/polyphagia, weight gain, polydipsia/polyuria) was performed based on historical information obtained from owners, referring veterinarians, and a review of patient medical records. Clinical signs of poldypsia/polyuria (PU/PD), polyphagia (PP), and weight gain (WG) were observed in 23.5% of patients (4 of 17; Table 2). Evaluation of serum chemistry revealed elevation of serum alkaline phosphatase (SAP) and/or alanine aminotransferase (ALT) in 47% of patients (8 of 17).
Serum analysis of IMR-CAR patients revealed elevated cholesterol levels in 23.5% of patients (4 of 17). Urine analysis revealed proteinuria/microalbuminuria in 77.7% (7 of 9; urine analysis data were missing for eight patients; Table 2). History of pancreatitis with abnormal cPL or elevated sPL, or elevated serum lipase was present in 30% (3 of 10; this was not evaluated in 7). Serum calcium was elevated in 5% of patients (1 of 17). Blood pressure evaluation revealed systemic hypertension in 20% of patients (2 of 10; this was not evaluated in 7; Table 2). Systemic hypertension was classified as systolic blood pressure equal or higher than 160 mm Hg. Historical presence of allergic and autoimmune diseases was documented for 70.5% of patients (12 of 17, Table 2). Atopy (91.6%, 11 of 12), food allergies (16.6%, 2 of 12), and dry eye disease (16.6%, 2 of 12) were the most frequently observed allergic/autoimmune diseases ( Table 2).

CAR retinas
In order to characterize the presence of immune cells in retinal tissue, antibodies against T-cell marker (anti-CD3), B-cell marker (anti-CD79), microglial marker (CD11b), and plasma-cell marker (anti-Ig cocktail) were used. Immunostaining analysis showed focal presence of immune cells (Figure 6), which were usually located in perivascular spaces and regions of focal retinal damage. T cells seemed to be the predominant cell population identified in retinas of deceased patients (Table 4). We did not perform IHC grading of evaluated retinas, since retinal sections frequently had just a few focal regions of cellular infiltrates per evaluated slide.  Table 1, patient no. 1); (C) and (D) poor but present PLR after red light illumination and good PLR after blue light illumination (Miniature Pinscher, CM, 7 y old, Table 1

| Retina serum autoantibodies-Western blot analysis
Western blotting analysis revealed the presence of serum anti-retinal autoAbs in 80% of tested canine IMR-CAR patients (Table 5, Figure 7). Serum samples of glaucomatous dogs were positive for autoAbs in 100% of cases, while 70% of control healthy dogs were seropositive for anti-retinal autoAbs. Analysis of the antibody reactivity revealed similarities in retinal antigen recognitions between serum samples of healthy controls, glaucoma dogs, and IMR-CAR dogs.

CAR retina
Gene expression analysis was carried out using retinal tissue from two eyes of one dog with untreated IMR-CAR (CAR, patient 2, Table 1), two eyes from one IMR-CAR dog who had received IVIg treatment (CAR+, patient 1, Table 1), two eyes from healthy control dogs (Control), and three eyes from three dogs with a diagnosis of SARDS (SARDS). After normalization and removal of probesets with low expression (log2 < 6) in all samples, 23 286 probesets remained for analysis. To evaluate the global relationship of the samples to one another, the 2500 probesets exhibiting the highest variability across all samples were identified and used for principal component analysis (PCA) while, due to the limitations of the software, the highest 1200 were used for hierarchical clustering. The results of the PCA ( Figure 8A) demonstrate a close molecular relationship between the control samples and CAR+, concordant with the treatment success in these animals. CAR samples are located close together and near the controls, separated from them mostly due to differences in the lesser PC2 vector, indicating a relatively close relationship of the gene expression profiles. Retinal expression of the SARDS eye exhibited substantial variability between samples, but all were clearly distinct from both the CAR and control samples. These relationships were further confirmed using hierarchical clustering ( Figure 8B). Here, the two tightest clusters were formed between the two CAR and CAR+ samples, which were then clustered with the control. Again, SARDS samples formed a distinct group, clustered separately from all other samples.
We then compared gene expression directly between control, CAR, and CAR+ samples (Tables 6, 7 and 8). Applying a signal cut-off of log2 > 6 in at least one sample, statistical significance <0.05, and a twofold expression change identified 107 probesets representing 83 genes with elevated  Table 1, patient no. 6). F, Healthy (control) canine retina shows sporadic presence of the quiescent microglial cell (round cell with regular margins) labeled with anti-CD11b antibody (white arrow); (G) Canine CAR-IMR retina shows more prominent microglial cell numbers and phenotype (cells with irregular margins marked by white arrows) in the inner and outer plexiform layers, and outer nuclear layer (Miniature Pinscher (CM, 7 y old, Table 1, patient no. 6)). Legend: ONL, outer nuclear layer; INL, inner nuclear layer Note: CD3 + T cells were the most frequently observed cell population infiltrating CAR canine retinas (freshly collected ocular tissue was available from dogs 1, 2, and 6- Table 1). We show a portion of all detected changes along with functional classifications. Expression values represent log2 of detected signal strength. Note that one gene may be assigned to more than one functional category, but is only listed here once. One asterisk behind the probeset ID indicates that two probesets for the same gene were detected. Two asterisks indicate that at least three probesets were detected for this gene.
T A B L E 6 (Continued) increased transcription of immunoglobulins in CAR eyes. These transcripts were generally absent in control eyes but were expressed at a high level in the disease. We also detected increased expression of genes related to cancer signaling pathways, in particular those belonging to the PI3K-Akt pathway, as well as a small number of chemokines. In addition, increased expression of metabolic pathways and ER or Golgi components might have been indicative of a heightened metabolic demand in these eyes. Notably absent in among these genes were those typically associated with inflammatory cells, suggesting that only a small number of leukocytes were present in the retinas, that is, those indicative of active inflammation of the tissue. Reduced expression was observed for a largely disparate group of genes, although expression changes were typically mild. However, decreased expression of a number of genes containing an epidermal growth factor (EGF) domain was noted, as were molecules limiting PI3K-Akt pathway signaling.
When the gene expression patterns of the CAR samples were compared to those from SARDS eyes, we detected 254 probesets representing 199 genes with a higher expression in CAR, and 167 probesets of 136 genes more prominently expressed in SARDS retinas. As noted above, CAR samples expressed a substantial amount of immunoglobulins. As in our previous study, mRNA encoding immunoglobulins was also present in SARDS eyes, but to a significantly smaller extent than in CAR. 5 Additionally, we observed increased transcripts related to mitogen-activated protein kinase (MAPK) and PI3K-Akt signaling, and elevated expression of a number of genes that have been associated with the development of cancer. Finally, a number of transcriptional changes indicated that CAR samples had increased metabolic demand.
In contrast, SARDS eyes displayed significantly higher expression of proinflammatory molecules, including IL-18, TLR1, and chemokines and their receptors (Table 8). We also detected higher transcriptional activity for gene products related to antigen presentation, primarily related to MHC class II. Finally, significantly higher expression of genes associated with T, B, or NK cells was observed in the SARDS samples compared to CAR. The presence of these transcripts most likely indicates that a larger number of these cells infiltrate into the SARDS retina than the CAR. Note: A portion of all detected changes are shown along with functional classifications. Expression values represent log2 of detected signal strength. Note that one gene may be assigned to more than one functional category, but is listed only once here. One asterisk behind the probeset ID indicates that two probesets for the same gene were detected. Two asterisks indicate that at least three probesets were detected for this gene.

IMR patients
Immunosuppressive therapy resulted in the reversal of complete blindness in 44% (4 of 9) of treated patients, with 61% of all treated patients recovering and/or maintaining vision (8 of 13, Table 3). Median time for dogs maintaining vision was five months (range 1-35 months), which corresponded to survival time. Initially, 53% (9 of 17) patients were treated with high doses of systemic steroids (prednisone: n = 6; prednisone + doxycycline: n = 3), with only 22% (2 of 9) responding to therapy, while 78% (7 of 9) did not respond. A total of four nonresponding patients were treated with the intravitreal IVIg injection, and three of those (75%) responded with the recovery of vision and improvement of cPLR and retinal electrical responses (Figures 9 and 10).

| DISCUSSION
Human AIR (npAIR and CAR) are relatively rare autoimmune forms of diseases characterized by progressive loss of vision, severe depression of retinal electrical activity, minimal fundus changes, and presence of retinal autoAbs in the serum samples of human patients. 27,28 CAR was first reported in human patients almost five decades ago; however, reports of the CAR in the veterinary patient population have not been previously published in the peer-reviewed literature. 41 Grozdanic et al (2008) introduced a new classification of the possible AIR in dogs, proposing a classification separating a nonparaneoplastic clinical entity (SARDS), and IMR, in which some patients fit the criteria of CAR, while the majority did not have evidence of cancer. 2 Sudden acquired retinal degeneration syndrome was first reported almost four decades ago, and in recent years, there has been sufficient evidence characterizing it as an immune-mediated type of retinopathy. 5,6 Blindness caused by SARDS is most frequently seen in middle-aged/older females of small breeds, with mixed-breed dogs, Dachshunds, Pugs, and Miniature Schnauzers being the most frequently affected. 42 In this study, there was an equal distribution between the large and small breed dogs affected, with males being more prevalent compared to females.
Sudden acquired retinal degeneration syndrome diagnostic features have been previously described: sudden onset of blindness, completely extinguished retinal electrical responses, absent chromatic PLR after red light illumination, normal chromatic PLR after the blue light illumination, and near-normal fundus appearance. 2,5,34 In this study, none of the patients evaluated satisfied previously established criteria for SARDS diagnosis, and we have tried to describe critical clinical features which may help differentiating CAR and SARDS patients (Table 9). While seven had completely extinguished full-field electrical responses when combined rod-cone evaluation ERG routine was used, none had the characteristic chromatic PLR deficits previously described in SARDS patients (no red-good blue response). We detected a variety of retinal electrical responses in rest of the patients, which again is not a feature previously reported in SARDS patients. Observed combination of cPLR and electrophysiological changes in this patient population corresponded to the features previously described in IMR patients. 2 Detailed fundus evaluation revealed numerous funduscopic abnormalities, very similar to those observed in SARDS eyes and described in a recent paper. Retinal vascular attenuation was the most frequently observed. 5 Presence of frequently observed retinal inflammatory lesions such as perivascular hyper-reflective lesions, perivascular retinal edema, and perivascular exudative lesions potentially support the notion of the possible inflammatory nature of the disease, and identical findings have been recently described in SARDS retinas. 5 This study effectively utilized SD-OCT technology to demonstrate presence of many retinal lesions (loss of the Note that one gene may be assigned to more than one functional category, but is listed only once. One asterisk behind a probeset ID indicates that two probesets for the same gene were detected. Two asterisks indicate that at least three probesets were detected for this gene. inner and outer segment photoreceptor stratification, loss of the outer segments, loss of the outer nuclear layer, retinal detachment, focal retinal thinning, and chorioretinal scars) previously reported in SARDS eyes, which were not easily detected by indirect ophthalmoscopy fundus evaluation or that of retinal fundus images. 5 Careful analysis of retinal OCT scans revealed predominantly perivascular localization of lesions with inflammatory appearance, which was also confirmed by histopathological analysis of limited retinal tissue specimens. An equally important finding of this study was the existence of the loss of the inner segment-outer segment (IS-OS) photoreceptor delineation, and presence of photoreceptor loss, which was particularly prominent in perivascular regions and frequently had a focal appearance. Identical findings have been previously described as the most frequently observed features in human npAIR and CAR patients. 43,44 Considering that all patients evaluated in this study had received a diagnosis of cancer, and many functional, structural, and funduscopic features had been previously observed in human CAR patients, we hypothesized that observed changes are indeed a result of the paraneoplastic retinal syndrome. However, we cannot exclude the possibility that these patients had an immune-mediated form of retinal disease independent of the concurrent cancer presence.
In previous publications, it has been demonstrated histological evidence of perivascular T-and B-cell inflammatory exudates, and MA evidence of inflammatory gene profile changes and primary photoreceptor lesions in perivascular regions observed by OCT imaging in SARDS retinas. 2,5 Similar findings were confirmed in this study to some extent, further reinforcing the notion of possible immune-mediated events being responsible for observed functional and structural retinal changes. This is further supported by the fact that systemic and intraocular immunosuppressive and immunomodulatory therapy resulted in the restoration and preservation of vision in the majority of treated patients, as is frequently case in human CAR. 45 Previous studies found a relatively high incidence of metabolic abnormalities in SARDS patients, such as PU and PD (38%, 42 30% 1 ), PP (19%, 42 20% 1 ), and SAP elevation (37%, 46 28% 1 ), which were in good accord with our findings in this study. This potentially indicates a very similar nature of systemic changes when comparing SARDS and CAR patients. This study confirmed very high incidence of proteinuria (75%), which was similar to the recent data from SARDS patients in Canada, but much higher than previously published studies from the SARDS patient population in the United States. 5,46,47 In this study, we attempted to characterize molecular events associated with CAR retinal damage in dogs, with the goal of better understanding the pattern of observed morphological and functional changes. This report is the first to describe detailed changes in the gene expression pattern and potential immunological consequences in species with large eyes, and spontaneously occurring disease similar to CAR in humans. A major limitation of the study was the small number of CAR samples analyzed. While this likely influenced F I G U R E 9 ERG tracings in CAR-treated canine patients. (A1-A2) Patient 1, Table 1, meningioma: ERG testing revealed severe decrease in full-field scotopic ERG amplitudes and flicker ERG amplitudes with development of blindness despite very high dose of systemic steroids. Complete recovery of vision in photopic conditions with significant improvement in retinal electrical activity was detected 14 d after intraocular IVIg injection. Dog remained blind in scotopic conditions. (B1-B5) ERG tracings from the CAR patient (Patient 16, Table 1) with sarcoma before and after intravitreal IVIg therapy: (B1) Full-field ERG tracings showed completely absent retinal electrical activity in both eyes (the pretreatment recording was done by primary ophthalmologist using the Retinographics full-field combined rod-cone routine; no other ERG testing routines were pursued). Patient was completely blind at presentation despite treatment with immunosuppressive dose of systemic steroids; (B2) rod response and full-field scotopic combined rod-cone response showed mild recovery in retinal electrical function 60 d after intravitreal IVIg injection; (B3) oscillatory potentials (op) were not recordable at the same time point; (B4) photopic cone and flicker response (B5) recordings showed recovery of cone electrical activity. Patient remained visual in photopic and mesopic conditions for 12 mo post-treatment. (C1-C3) ERG tracings from the CAR patient (Patient 3, Table 1) with meningioma: (C1) Full-field ERG tracings showed by primary ophthalmologist revealed dramatic reduction in ERG amplitudes (b-wave = 22 µV) and absent cPLR after red light illumination with mildly decreased but present cPLR after the blue light illumination. Systemic steroids were initiated. (C2) Retinal electrical evaluation performed 7 d after the initiation of steroid therapy revealed normalization of responses (b-wave amplitude = 130 µV, recordings performed at our institution). Vision was significantly improved. Twelve months after initiating treatment, sudden onset of complete vision loss was detected and ERG evaluation revealed 33 µV b-wave amplitudes (ERG tracings not available, performed by primary ophthalmologist), and patient was referred to us for intravitreal IVIg treatment resulting in significant vision improvement (object tracking capability in day light conditions was restored). (C3) Completely normalized retinal electrical activity is present 22 mo after IVIg treatment; however, owner reports development of intermittent episodes of decreased vision, which correspond to episodes of pancreatitis flare ups. Due to the pancreatitis episodes and dramatic liver enzyme elevations, systemic steroids were first substituted with systemic cyclosporine (5 mg/kg BW q24h PO), which was not tolerated. Systemic cyclosporine was substituted with leflunomide (4 mg/kg BW q24h PO), which was tolerated for the period of 6 mo, and had to be discontinued. At that time, therapy had to be continued with subconjunctival steroid injections (monthly) and intravitreal steroid injections (once every 6 mo) combined with the low dose of systemic steroids (0.5 mg/kg BW q48h PO). Only full-field combined rod-cone routines were done for this patient during all examination time points when ERG testing was pursued. Legend: BW-body weight the expression levels of some genes, the overall similarity between immunoglobulin gene expression in this study and our previously reported data from SARDS retina suggest that observed gene expression changes may indeed reflect similarities in the immunological reaction, with SARDS immunological profile being more aggressive compared to CAR. 5 As a consequence of the small sample size, expression changes of individual genes should be regarded with care but more confidence can be placed in the identified pathways as these are supported by multiple genes. Furthermore, while it is conceivable that the eyes used for gene expression analysis represented atypical cases, this concern is alleviated by the overall similarity between immunoglobulin gene expression in this study and our previously reported SARDS data. 5 Interestingly, molecular profiling and gene expression map comparison between SARDS and CAR eyes showed distinct differences in overall gene expression despite almost identical clinical presentation in many cases; however, difference in therapeutic response (60% in this study compared to 35%-40% for blind SARDS patients, Table 9) should serve as an additional evidence of the more aggressive nature of disease in SARDS patients.
Among genes with elevated expression levels in CAR retinas, genes mediating various aspects of immune-mediated response were by far the most prevalent. Prominent functional categories of genes with elevated expression in CAR retinas included antigen presentation, leukocyte activation and adhesion, lysosomal and proteasome activity, and immunoglobulin production. In addition, numerous genes with a function in apoptosis and inflammation signaling were more abundant in CAR retinas. Our analyses also indicated that CAR led to lower expression levels for a large number of genes, primarily associated with photoreceptor function. Reduced mRNA levels of individual genes could partially be the result of transcriptional control mechanisms; however, a more likely explanation is that observed changes were the result of the primary photoreceptor damage and apoptotic loss in CAR eyes, further supporting the OCT and histology data presented here. These findings are consistent with the previous report by Grozdanic et al, who showed similar molecular and functional changes in SARDS retinas. 5 Interestingly, while strong upregulation of different complement genes has been described in SARDS retinas, in this case the situation was not similar in CAR eyes. 5 Observed histological and molecular changes in CAR retinas provide a unique opportunity to study numerous possible immune-mediated responses against cancer without the background molecular noise that comes with the rapid proliferation of neoplastic cells and secondary inflammatory responses associated with neoplasia-induced tissue necrosis and destruction. Observed immunological mechanisms in CAR retinas recapitulate some previously reported mechanisms activated F I G U R E 1 0 cPLR testing in CAR patient before and after intraocular IVIg treatment (Patient 1, Table 1, meningioma). (A) Resting pupil diameter in photopic conditions before IVIg treatment; (B) red light illumination did not elicit pupil light reflex before IVIg treatment; (C) recovery of the PLR after red light illumination was detectable 14 d after intravitreal IVIg injection T A B L E 9 Proposed classification of autoimmune retinopathies (AIR) in canines based on the clinical presentation and currently established human nomenclature