Introduction

Over the past 2 decades, veterinarians in North America have increasingly recognized CuAH in pet dogs.^1,2,3,4 This phenomenon seemingly corresponds with the adoption in 1997 of new guidelines for copper content in commercial dog foods.^a At that time, concerns had been raised regarding the low bioavailability of food-grade copper oxide as a source of dietary copper,⁵ and on the basis of results of a small study^a involving juvenile dogs, a recommendation was made to replace copper oxide in commercial dog foods with bioavailable forms of copper, even though there was no evidence of widespread copper deficiency in pet dogs. Subsequent to adoption of these new guidelines, 2 studies^1,2 documented increases over time in hepatic copper concentrations in breeds considered to be predisposed to develop CuAH (eg, Labrador Retrievers) and in breeds not considered to be predisposed to this condition. Another study⁶ demonstrated high copper concentrations in some brands of commercial dog foods.

Hepatic lesions associated with CuAH are fully reversible when appropriate treatment is administered, especially if treatment is initiated early in the disease process. As a result, veterinarians are now more inclined to recommend hepatic biopsy for dogs with chronically high hepatic enzyme activities, particularly alanine aminotransferase activity, and in dogs in which hepatic enzyme activities repeatedly fluctuate.

Hepatic biopsy is the only procedure that can confirm the nature and severity of hepatic disease. In particular, making a diagnosis of CuAH requires histologic confirmation of pathological copper accumulation in hepatic specimens stained with a copper-specific stain and documentation of a high hepatic copper concentration.^7,8 Historically, AAS has been used to quantify hepatic copper concentration; however, other methods have been used, including ICP-MS and digital image analysis of hepatic sections stained with rhodanine.⁹

Because of the equipment and technical expertise required, spectroscopic-spectrometric methods of quantifying hepatic copper concentrations (ie, AAS and ICP-MS) are typically performed at large diagnostic laboratories. Most often, when multiple hepatic biopsy specimens are submitted, a single specimen will be randomly selected for spectroscopic-spectrometric testing. However, previous studies^{10,11,12,13,14,15,16} involving human patients have documented uneven hepatic copper distribution, particularly in patients with remodeling (regenerative nodules and fibrosis) due to chronic liver disease. Notably, this phenomenon complicates histologic detection of copper-storage hepatopathy (Wilson disease) in humans.^14,15,16 Additional studies^13,16 involving human patients confirm that small biopsy specimen size and analysis of a single biopsy specimen have a negative impact on the ability to detect pathological hepatic copper accumulation.

Pathological hepatic copper accumulation in dogs is easily recognized histologically with copper-specific stains if representative tissue samples are submitted.^7,8 The authors prefer the use of rhodanine stain for this purpose, because it vividly demonstrates copper-protein aggregates as bright orange-red cytosolic granules. One shortcoming of rhodanine stain is that the color fades over time, particularly when slides are exposed to light.⁸ However, this can be overcome by digitally scanning recently stained slides.

Rhodanine staining of hepatic specimens allows assignment of a qualitative score for severity of copper accumulation. In addition, digital image analysis of rhodanine-stained hepatic sections has been used to quantify copper content in the section and can be used to provide an estimate of overall mean hepatic copper concentration. Digital image analysis as a method of quantifying hepatic copper concentration has been validated for use in dogs, cats, and ferrets with homogeneous hepatic copper distributions.^9,17,b

Over the years, one of the authors (SAC) has encountered a concerning number of cases involving dogs for which spectroscopic-spectrometric (AAS or ICP-MS) measurements of hepatic copper concentrations differed substantially from concentrations expected on the basis of qualitative copper accumulation scores assigned to rhodanine-stained hepatic specimens and from estimated hepatic copper concentrations determined by means of digital image analysis of rhodanine-stained hepatic sections. We hypothesize that this discordance reflects uneven tissue copper distribution, potentially as a result of hepatic remodeling (eg, development of fibrosis, regenerative nodules, or parenchymal extinction); evaluation of single rather than multiple tissue sections; or variations in tissue moisture content, because spectroscopic-spectrometric hepatic copper concentrations are reported on a dry-weight basis.

The study reported here was designed to further investigate disparities in hepatic copper concentrations determined by spectroscopic-spectrometric methods and digital image analysis and to compare qualitative copper accumulation scores with hepatic copper concentrations determined with these various methods. Specifically, we wanted to determine the concordance between hepatic copper concentrations determined by spectroscopic-spectrometric methods (AAS or ICP-MS) and digital image analysis for hepatic biopsy specimens from dogs with a variety of hepatic disorders, determine whether uneven hepatic copper distribution could explain any discordance between hepatic copper concentrations determined with spectroscopic-spectrometric methods versus digital image analysis, evaluate the utility of hepatic copper accumulation scores assigned to rhodanine-stained hepatic sections in predicting whether spectroscopic-spectrometric hepatic copper concentrations might be misleading, identify histologic features significantly associated with pathological hepatic copper accumulation, and determine whether body weight, age, or sex of dogs with various hepatic disorders was significantly associated with hepatic copper concentration.

Materials and Methods

Logbooks for one of the authors (SAC) were searched to identify dogs for which hepatic biopsy specimens had been submitted between January 1999 and December 2019 for evaluation of copper content. Dogs were included in the study if hepatic copper concentration had been quantified by means of AAS or ICP-MS and sections had been examined histologically for evidence of pathological copper accumulation.

Medical records of dogs included in the study were retrieved, and information on signalment (breed, age, sex, and body weight), biopsy method (surgical wedge, laparoscopic cup, or 14-gauge needle), and number of sections available for rhodanine staining were recorded. Dogs for which needle biopsy specimens had been submitted were included only if ≥ 3 core samples provided a total of ≥ 15 portal tracts.¹⁸

All biopsy specimens were fixed in neutral-buffered 10% formalin, sectioned at a thickness of 7 μm, and routinely stained with H&E, Masson trichrome (for fibrillar collagen), and rhodanine (for copper) stains in the Diagnostic Histopathology Laboratory of the Cornell University College of Veterinary Medicine. Adequacy of rhodanine staining was verified in all cases by the inclusion of concurrent positive and negative control samples. All rhodanine-stained sections were digitally scanned,^c with images saved to a portable hard drive. For specimens for which rhodanine stains had faded, new sections were prepared for staining with rhodanine stain.

Histologic sections were initially examined by a board-certified veterinary pathologist and resident-intraining. Subsequently, all sections were independently evaluated by 2 of the authors (SAC and ADM) who both have extensive experience in hepatic histopathology.

For all histologic sections, the following 6 features were scored as present or absent: necroinflammatory disease, nonspecific inflammatory infiltrates in portal or centrilobular regions (in the absence of hepatitis), glycogen-type vacuolar hepatopathy, and changes consistent with portal venous hypoperfusion, ductal plate malformation, or toxic hepatopathy. Necroinflammatory disease was considered to be present if inflammatory infiltrates associated with necrotic hepatocytes were seen, inflammatory infiltrates in central or portal regions breached the margins of delimiting adventitia, inflammatory infiltrates coordinated with tissue remodeling were present, or evidence consistent with extrahepatic bile duct obstruction or cholangitis was observed. Nonspecific inflammatory infiltrates were defined by localization in portal tracts or within the adventitia of central regions, without evidence of a necroinflammatory impact on hepatic structure. Vacuolar hepatopathy was considered present if most hepatocytes had glycogen-type cytosolic expansion. Portal venous hypoperfusion was considered present if there was evidence of lobular atrophy (ie, abnormally close approximation of lobular elements, small hepatocytes), increased cross-sections of sinuous, thick muscular arteries in portal tracts, occasional arterial twigs in hepatic parenchyma unassociated with portal elements, or increased numbers of diminutive portal tracts with or without apparent portal vein silhouettes. Ductal plate malformation was defined by identification of malformed, proliferative-like bile ducts in portal tracts embedded in exuberant extracellular matrix and forming fibroductal trabeculae consistent with a proliferative-like ductal plate malformation phenotype, or by the presence of malformed ductal elements with dilated irregular silhouettes with variable budding protrusions and satellite bile duct elements consistent with the Caroli ductal plate malformation phenotype.¹⁹ Toxic hepatopathy was considered present if hepatocytes had microvesicular cytosolic lipid vacuolation, anisocytosis, megalohepatocytosis, anisokaryosis, or ring mitoses. The following 3 histologic changes reflective of architectural remodeling were also designated as present or absent: dissecting or bridging hepatic fibrosis, regenerative nodules, and parenchymal extinction (ie, remodeled regions devoid of viable hepatocytes). Rhodanine-stained sections were evaluated for the presence or absence of uneven copper distribution, and a qualitative hepatic copper accumulation score ranging from 0 (hepatocytes devoid of stainable cytosolic copper) to 5 (all hepatocytes display stainable cytosolic copper) was assigned on the basis of the percentage of centrilobular hepatocytes with cytosolic copper aggregates and extent of midzonal involvement versus panlobular distribution (Supplementary Figure S1, available at: avmajournals.avma.org/doi/suppl/10.2460/javma.258.4.395).

For all dogs included in the study, hepatic copper concentration had been measured by means of spectroscopic-spectrometric (AAS or ICP-MS) and digital image analysis. All concentrations were expressed as parts per million (μg/g) on a dry-weight basis. For all dogs, digital image analysis was performed by the individual who had developed and validated the methodology (SAC).⁹ Digital image analyses were performed on areas of interest for biopsy sections with viable hepatic parenchyma (ie, areas of extinct parenchyma and severe fibrosis were excluded) and were completed without knowledge of spectroscopic-spectrometric results. Biopsy specimens for which ≥ 4 rhodanine-stained sections were available had each section independently digitally analyzed to evaluate intraindividual variation in copper distribution across hepatic sections.

Results

A total of 516 dogs were included in the study. There were 257 males (22 sexually intact and 235 castrated) and 259 females (10 sexually intact and 249 spayed); the sex distribution was not significantly different from a 1:1 male-to-female distribution. Of the 516 dogs, 100 were mixed-breed dogs and 416 were purebred dogs. Purebred dogs consisted of 71 breeds; breeds represented by ≥ 10 dogs were Doberman Pinscher (n = 76), Labrador Retriever (43), Shih Tzu (28), Yorkshire Terrier (18), Miniature Schnauzer (17), Cocker Spaniel (16), Maltese (14), Shetland Sheepdog (12), Havanese (12), and Chihuahua (10). Of the 416 purebred dogs, 194 (47%) represented 35 breeds with typical adult weights ≥ 20 kg and 222 (53%) represented 36 breeds with typical adult weights < 20 kg. Median hepatic copper concentrations determined by means of AAS and digital image analysis were significantly higher for dogs representing breeds with a typical adult weight ≥ 20 kg than for dogs representing breeds with a typical adult weight < 20 kg (Table 1). However, median hepatic copper concentrations determined by means of ICP-MS were not significantly different between these 2 groups. To identify potential bias associated with the large numbers of Doberman Pinschers and Labrador Retrievers (breeds considered to be predisposed to CuAH), these analyses were repeated after Doberman Pinschers and Labrador Retrievers were removed from the group of dogs with typical adult weights ≥ 20 kg, and the findings were the same.

Hepatic copper concentrations were not significantly different between male and female dogs. Median hepatic copper concentration determined by means of digital image analysis was significantly higher for senior (≥ 9 years old) than for younger (< 9 years old) dogs (Table 1). However, median hepatic copper concentrations determined by means of AAS and median copper concentrations determined by means of ICP-MS were not significantly different between age groups.

Biopsy method was classified as surgical wedge in 191 (37%) dogs, laparoscopic cup in 121 (23%) dogs, and 14-gauge needle in 204 (40%) dogs. For 409 dogs, hepatic copper concentration was determined by both AAS and digital image analysis, and biopsy method for these dogs was classified as surgical wedge in 125 (31%) dogs, laparoscopic cup in 81 (20%) dogs, and 14-gauge needle in 203 (50%) dogs. For 108 dogs, hepatic copper concentration was determined by both ICP-MS and digital image analysis, and biopsy method for these dogs was classified as surgical wedge for 67 (62%) dogs, laparoscopic cup for 40 (37%) dogs, and 14-gauge needle for 1 (1%) dog. In 1 dog, hepatic copper concentration was determined by both AAS and ICP-MS, in addition to digital image analysis. Number of hepatic biopsy specimens submitted per dog was significantly (P < 0.001) higher for dogs that underwent laparoscopic cup biopsy (median, 5 specimens/dog; range, 2 to 15 specimens/dog; 95% CI, 5 to 6 specimens/dog) than for dogs that underwent surgical biopsy (median, 3 specimens/dog; range, 1 to 9 specimens/dog; 95% CI, 2 to 3 specimens/dog) or 14-gauge needle biopsy (median, 3 specimens/dog; range, 3 to 5 specimens/dog; 95% CI, 3 to 3 specimens/dog).

For the 409 dogs with paired hepatic copper concentrations obtained with digital image analysis and AAS, median concentration obtained with digital image analysis (median, 511 ppm; range, 180 to 20,622 ppm; 95% CI, 1,123 to 1,554 ppm) was significantly (P < 0.001) higher than median concentration obtained with AAS (median, 386 ppm; range, 18 to 11,700 ppm; 95% CI, 677 to 897 ppm). Similarly, for the 108 dogs with paired hepatic copper concentrations obtained with digital image analysis and ICP-MS, median concentration obtained with digital image analysis (median, 621 ppm; range, 191 to 13,141 ppm; 95% CI, 1,017 to 1,752 ppm) was significantly (P < 0.001) higher than median concentration obtained with ICP-MS (median, 523 ppm; range, 7 to 6,642 ppm; 95% CI, 585 to 932 ppm). Similar findings were obtained when these analyses were repeated with dogs grouped on the basis of biopsy collection method. That is, for dogs that underwent surgical wedge biopsy, hepatic copper concentration obtained with digital image analysis was significantly higher than concentrations obtained with AAS (P < 0.001) or ICP-MS (P = 0.002); for dogs that underwent laparoscopic cup biopsy, hepatic copper concentration obtained with digital image analysis was significantly higher than concentrations obtained with AAS (P < 0.001) or ICP-MS (P < 0.001); and for dogs that underwent 14-gauge needle biopsy, hepatic copper concentration obtained with digital image analysis was significantly higher than concentration obtained with AAS (P < 0.001). Concentrations obtained with digital image analysis versus ICP-MS for samples obtained by 14-gauge needle biopsy were not analyzed, because only 1 sample fit these criteria.

Of the 516 dogs, 256 (50%) had histologic evidence of necroinflammatory disease, 165 (32%) had glycogen-type vacuolar hepatopathy, 124 (24%) had changes consistent with portal venous hypoperfusion, 105 (20%) had nonspecific inflammatory infiltrates, 30 (6%) had features consistent with ductal plate malformation, and 15 (3%) had evidence of toxic hepatopathy. Regardless of the analytic method (digital image analysis, AAS, or ICP-MS), dogs with necroinflammatory disease had significantly higher hepatic copper concentrations than did dogs without necroinflammatory disease (Table 2). For dogs with necroinflammatory disease, hepatic copper concentrations obtained with digital image analysis were significantly higher than concentrations obtained with AAS and ICP-MS. For dogs without necroinflammatory disease, hepatic copper concentrations obtained with digital image analysis were significantly higher than concentrations obtained with AAS but were not significantly different from concentrations obtained with ICP-MS. In dogs with glycogen-type vacuolar hepatopathy, portal venous hypoperfusion, nonspecific inflammatory infiltrates, ductal plate malformation, and toxic hepatopathy, hepatic copper concentrations obtained with digital image analysis were significantly higher than concentrations obtained with AAS. Finally, in dogs with glycogen-type vacuolar hepatopathy, but not in dogs with portal venous hypoperfusion, nonspecific inflammatory infiltrates, ductal plate malformation, or toxic hepatopathy, hepatic copper concentrations obtained with digital image analysis were significantly higher than concentrations obtained with ICP-MS.

For each of the 3 analytic methods (digital image analysis, AAS, and ICP-MS), hepatic copper concentration was significantly higher in dogs with uneven copper distribution (n = 139), hepatic fibrosis (169), regenerative nodules (121), or parenchymal extinction (73) than in dogs without these histologic features, with the exception that hepatic copper concentration obtained with ICP-MS did not differ between dogs with versus without parenchymal extinction (Table 3; Figures 1 and 2). Significant differences were also found between hepatic copper concentrations determined by digital image analysis versus spectroscopic-spectrometric methods among dogs with and without uneven tissue copper distribution and among dogs with and without fibrosis, regenerative nodules, and parenchymal extinction. On inspection, spectroscopic-spectrometric copper concentrations were consistently lower (34% to 63%) than concentrations determined with digital image analysis in dogs with uneven copper distribution, fibrosis, regenerative nodules, and parenchymal extinction. Comparatively, dogs lacking uneven hepatic copper distribution and these discerning histologic features had spectroscopic-spectrometric copper concentrations that more closely approximated concentrations obtained with digital image analysis (ie, 69% to 104%).

Photomicrographs of hepatic sections from a dog with uneven hepatic copper distribution. A—Low-magnification photomicrograph of 6 H&E-stained biopsy sections with a wide variation in hepatic copper concentration among sections. The 2 sections in the top left and the section in the bottom right have no remaining viable hepatic parenchyma (parenchymal extinction). Browntan material represents congested blood, macrophages laden with cellular debris and lipofuscin (oxidative membrane debris), and fibrosis. The top right section and the 2 sections in the bottom left show remodeled parenchyma with incompletely defined regenerative nodules. H&E stain; bar = 5 mm. B—Low-magnification photomicrograph of similar sections stained with Masson trichrome stain. The 2 sections in the top left and the section in the bottom right have a diffuse distribution of blue-staining fibrillar collagen. Sections with viable hepatic parenchyma are nodular and have areas with expansive fibrosis discernable even at this low magnification. Masson trichrome stain; bar = 5 mm. C—Low-magnification photomicrograph of similar sections stained with rhodanine stain. Sections with viable hepatic parenchyma have orange-red aggregates of copper in hepatocytes near developing nodules. Rhodanine stain; bar = 6 mm. Inset—High-magnification photomicrograph illustrating the severity of copper accumulation in viable parenchyma. Rhodanine stain; bar = 200 μm. Hepatic copper concentrations determined by digital image analysis were substantially lower in areas of extinct parenchyma (< 600 ppm) than in areas with viable hepatic parenchyma (2,458 to 3,117 ppm).

Citation: Journal of the American Veterinary Medical Association 258, 4; 10.2460/javma.258.4.395

• Download Figure
• Download Figure
• Download figure as PowerPoint slide

View Full Size

Photomicrographs of hepatic sections from a dog with uneven hepatic copper distribution. A—Low-magnification photomicrograph of 6 H&E-stained biopsy sections with a wide variation in hepatic copper concentration among sections. The 2 sections in the top left and the section in the bottom right have no remaining viable hepatic parenchyma (parenchymal extinction). Browntan material represents congested blood, macrophages laden with cellular debris and lipofuscin (oxidative membrane debris), and fibrosis. The top right section and the 2 sections in the bottom left show remodeled parenchyma with incompletely defined regenerative nodules. H&E stain; bar = 5 mm. B—Low-magnification photomicrograph of similar sections stained with Masson trichrome stain. The 2 sections in the top left and the section in the bottom right have a diffuse distribution of blue-staining fibrillar collagen. Sections with viable hepatic parenchyma are nodular and have areas with expansive fibrosis discernable even at this low magnification. Masson trichrome stain; bar = 5 mm. C—Low-magnification photomicrograph of similar sections stained with rhodanine stain. Sections with viable hepatic parenchyma have orange-red aggregates of copper in hepatocytes near developing nodules. Rhodanine stain; bar = 6 mm. Inset—High-magnification photomicrograph illustrating the severity of copper accumulation in viable parenchyma. Rhodanine stain; bar = 200 μm. Hepatic copper concentrations determined by digital image analysis were substantially lower in areas of extinct parenchyma (< 600 ppm) than in areas with viable hepatic parenchyma (2,458 to 3,117 ppm).

Citation: Journal of the American Veterinary Medical Association 258, 4; 10.2460/javma.258.4.395

• Download Figure
• Download Figure
• Download figure as PowerPoint slide

Photomicrographs of hepatic sections from a dog with uneven hepatic copper distribution. A—Low-magnification photomicrograph of 6 H&E-stained biopsy sections with a wide variation in hepatic copper concentration among sections. The 2 sections in the top left and the section in the bottom right have no remaining viable hepatic parenchyma (parenchymal extinction). Browntan material represents congested blood, macrophages laden with cellular debris and lipofuscin (oxidative membrane debris), and fibrosis. The top right section and the 2 sections in the bottom left show remodeled parenchyma with incompletely defined regenerative nodules. H&E stain; bar = 5 mm. B—Low-magnification photomicrograph of similar sections stained with Masson trichrome stain. The 2 sections in the top left and the section in the bottom right have a diffuse distribution of blue-staining fibrillar collagen. Sections with viable hepatic parenchyma are nodular and have areas with expansive fibrosis discernable even at this low magnification. Masson trichrome stain; bar = 5 mm. C—Low-magnification photomicrograph of similar sections stained with rhodanine stain. Sections with viable hepatic parenchyma have orange-red aggregates of copper in hepatocytes near developing nodules. Rhodanine stain; bar = 6 mm. Inset—High-magnification photomicrograph illustrating the severity of copper accumulation in viable parenchyma. Rhodanine stain; bar = 200 μm. Hepatic copper concentrations determined by digital image analysis were substantially lower in areas of extinct parenchyma (< 600 ppm) than in areas with viable hepatic parenchyma (2,458 to 3,117 ppm).

Citation: Journal of the American Veterinary Medical Association 258, 4; 10.2460/javma.258.4.395

• Download Figure
• Download Figure
• Download figure as PowerPoint slide

View Full Size

Photomicrographs of hepatic sections from a dog with uneven hepatic copper distribution. A—Low-magnification photomicrograph of 6 H&E-stained biopsy sections with a wide variation in hepatic copper concentration among sections. The 2 sections in the top left and the section in the bottom right have no remaining viable hepatic parenchyma (parenchymal extinction). Browntan material represents congested blood, macrophages laden with cellular debris and lipofuscin (oxidative membrane debris), and fibrosis. The top right section and the 2 sections in the bottom left show remodeled parenchyma with incompletely defined regenerative nodules. H&E stain; bar = 5 mm. B—Low-magnification photomicrograph of similar sections stained with Masson trichrome stain. The 2 sections in the top left and the section in the bottom right have a diffuse distribution of blue-staining fibrillar collagen. Sections with viable hepatic parenchyma are nodular and have areas with expansive fibrosis discernable even at this low magnification. Masson trichrome stain; bar = 5 mm. C—Low-magnification photomicrograph of similar sections stained with rhodanine stain. Sections with viable hepatic parenchyma have orange-red aggregates of copper in hepatocytes near developing nodules. Rhodanine stain; bar = 6 mm. Inset—High-magnification photomicrograph illustrating the severity of copper accumulation in viable parenchyma. Rhodanine stain; bar = 200 μm. Hepatic copper concentrations determined by digital image analysis were substantially lower in areas of extinct parenchyma (< 600 ppm) than in areas with viable hepatic parenchyma (2,458 to 3,117 ppm).

Citation: Journal of the American Veterinary Medical Association 258, 4; 10.2460/javma.258.4.395

• Download Figure
• Download Figure
• Download figure as PowerPoint slide

Photomicrographs of hepatic sections from a dog with uneven hepatic copper distribution. A—Low-magnification photomicrograph of 6 H&E-stained biopsy sections with a wide variation in hepatic copper concentration among sections. The 2 sections in the top left and the section in the bottom right have no remaining viable hepatic parenchyma (parenchymal extinction). Browntan material represents congested blood, macrophages laden with cellular debris and lipofuscin (oxidative membrane debris), and fibrosis. The top right section and the 2 sections in the bottom left show remodeled parenchyma with incompletely defined regenerative nodules. H&E stain; bar = 5 mm. B—Low-magnification photomicrograph of similar sections stained with Masson trichrome stain. The 2 sections in the top left and the section in the bottom right have a diffuse distribution of blue-staining fibrillar collagen. Sections with viable hepatic parenchyma are nodular and have areas with expansive fibrosis discernable even at this low magnification. Masson trichrome stain; bar = 5 mm. C—Low-magnification photomicrograph of similar sections stained with rhodanine stain. Sections with viable hepatic parenchyma have orange-red aggregates of copper in hepatocytes near developing nodules. Rhodanine stain; bar = 6 mm. Inset—High-magnification photomicrograph illustrating the severity of copper accumulation in viable parenchyma. Rhodanine stain; bar = 200 μm. Hepatic copper concentrations determined by digital image analysis were substantially lower in areas of extinct parenchyma (< 600 ppm) than in areas with viable hepatic parenchyma (2,458 to 3,117 ppm).

Citation: Journal of the American Veterinary Medical Association 258, 4; 10.2460/javma.258.4.395

• Download Figure
• Download Figure
• Download figure as PowerPoint slide

Photomicrograph of hepatic sections from a dog with severely uneven hepatic copper distribution. Macrophages and hepatocytes laden with orange-red copper aggregates surround regenerative nodules, producing a mottled appearance, and regenerative nodules are separated by coalescing regions of centrilobular parenchymal collapse and inflammation. Areas with strong rhodanine staining had a copper concentration determined by digital image analysis as high as 11,906 ppm, whereas areas with only faint rhodanine staining had copper concentrations < 600 ppm. Rhodanine stain; bar = 5 mm.

Citation: Journal of the American Veterinary Medical Association 258, 4; 10.2460/javma.258.4.395

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Photomicrograph of hepatic sections from a dog with severely uneven hepatic copper distribution. Macrophages and hepatocytes laden with orange-red copper aggregates surround regenerative nodules, producing a mottled appearance, and regenerative nodules are separated by coalescing regions of centrilobular parenchymal collapse and inflammation. Areas with strong rhodanine staining had a copper concentration determined by digital image analysis as high as 11,906 ppm, whereas areas with only faint rhodanine staining had copper concentrations < 600 ppm. Rhodanine stain; bar = 5 mm.

Citation: Journal of the American Veterinary Medical Association 258, 4; 10.2460/javma.258.4.395

• Download Figure
• Download figure as PowerPoint slide

Photomicrograph of hepatic sections from a dog with severely uneven hepatic copper distribution. Macrophages and hepatocytes laden with orange-red copper aggregates surround regenerative nodules, producing a mottled appearance, and regenerative nodules are separated by coalescing regions of centrilobular parenchymal collapse and inflammation. Areas with strong rhodanine staining had a copper concentration determined by digital image analysis as high as 11,906 ppm, whereas areas with only faint rhodanine staining had copper concentrations < 600 ppm. Rhodanine stain; bar = 5 mm.

Citation: Journal of the American Veterinary Medical Association 258, 4; 10.2460/javma.258.4.395

• Download Figure
• Download figure as PowerPoint slide

For the 409 dogs with paired hepatic copper concentrations obtained with digital image analysis and AAS, the concentration obtained with digital image analysis exceeded the concentration obtained with AAS by ≥ 400 ppm in 114 (28%) dogs, by ≥ 600 ppm in 91 (22%) dogs, and by ≥ 1,000 ppm in 60 (15%) dogs, whereas the concentration obtained with AAS exceeded the concentration obtained with digital image analysis by ≥ 400 ppm in only 12 (3%) dogs. For the 108 dogs with paired hepatic copper concentrations obtained with digital image analysis and ICP-MS, the concentration obtained with digital image analysis exceeded the concentration obtained with ICP-MS by ≥ 400 ppm in 34 (31%) dogs, by ≥ 600 ppm in 27 (25%) dogs, and by ≥ 1,000 ppm in 19 (18%) dogs, whereas the concentration obtained with ICP-MS exceeded the concentration obtained with digital image analysis by ≥ 400 ppm in only 3 (3%) dogs.

For all 3 analytic methods, qualitative copper accumulation scores were significantly (all P < 0.001) associated with hepatic copper concentrations; however, the correlation was higher between qualitative score and concentration obtained with digital image analysis (r_s = 0.94) than between qualitative score and concentration obtained with AAS (r_s = 0.75) or ICP-MS (r_s = 0.57; Table 4). Even after removing dogs with hepatic copper concentrations obtained with digital image analysis ≥ 1,500 ppm, qualitative copper accumulation scores were significantly (all P < 0.001) associated with hepatic copper concentrations for all 3 analytic methods, but again, the correlation was higher between qualitative score and concentration obtained with digital image analysis (r_s = 0.92) than between qualitative score and concentration obtained with AAS (r_s = 0.67) or ICP-MS (r_s = 0.54).

For 134 dogs, ≥ 4 rhodanine-stained hepatic sections (total, 818 sections) were available for evaluation of intraindividual variation in copper distribution across hepatic sections (Table 5). The CV for hepatic copper concentrations obtained with digital image analysis was significantly higher for dogs with hepatic copper concentrations ≥ 1,000 ppm (P = 0.01) than for dogs with concentrations < 400 ppm and for dogs with hepatic copper concentrations ≥ 2,000 ppm (P = 0.02) than for dogs with concentrations < 400 ppm.

Discussion

Results of the present study indicated that in dogs, median hepatic copper concentrations obtained with spectroscopic-spectrometric methods (AAS and ICP-MS) were significantly lower than the median concentration obtained with digital image analysis of rhodanine-stained hepatic sections; that uneven hepatic copper distribution resulted in higher median hepatic copper concentration with digital analysis of rhodanine-stained sections, compared with AAS and ICP-MS methods; that qualitative copper accumulation scores correlated with measured hepatic copper concentrations; and that median hepatic copper concentration was significantly higher in larger dogs (ie, dogs of breeds with a typical adult weight ≥ 20 kg) than in smaller dogs, in older dogs (≥ 9 years) than in younger dogs, and in dogs with histologic evidence of necroinflammatory disease than in dogs without.

In the past, hepatic copper concentrations were typically measured by means of spectroscopic-spectrometric analysis (most commonly, AAS or ICP-MS) of randomly selected portions of unfixed or formalin-fixed hepatic specimens. Specimens analyzed by means of AAS are desiccated before flame photometry, whereas for specimens analyzed by means of ICP-MS, tissue moisture content is estimated. Spectroscopic-spectrometric analyses of hepatic copper concentration require a minimum of 50 mg of tissue, or approximately 10 mg of dry weight.^3,8,22 Although formalin-fixed, paraffin-embedded tissue can be used, this requires xylene extraction of paraffin and may result in loss of some tissue. It is essential that hepatic copper concentrations are normalized to a dry-weight basis to allow relevant comparisons to reference intervals. Failure of sample desiccation or underestimation of moisture content can result in erroneously low hepatic copper concentrations when measured by means of AAS or ICP-MS, respectively. The use of digital image analysis of rhodanine-stained hepatic sections and qualitative scoring of hepatic copper accumulation in the authors' hospital has disclosed discordance between spectroscopic-spectrometric results and results of examination of rhodanine-stained sections in some dogs, prompting the present study.

Digital image analysis of rhodanine-stained sections allows a critical assessment of copper distribution across all hepatic sections from an individual patient. A similar approach is impossible with spectroscopic-spectrometric methods, as it would exhaust biopsy material. In the present study, the hepatic copper concentration obtained with digital image analysis exceeded the concentration obtained with AAS by ≥ 400 ppm in 28% (114/409) of dogs and exceeded the concentration obtained with ICP-MS by ≥ 400 ppm in 31% (34/108) of dogs, whereas the concentration obtained with AAS or ICP-MS exceeded the concentration obtained with digital image analysis in only 3% (12/409 and 3/108, respectively) of dogs. We suspect that uneven copper distribution may have resulted in an underestimation of hepatic copper concentration with AAS and ICP-MS. Our findings were consistent with previously reported findings for human patients with Wilson disease.¹⁵ The fact that AAS and ICP-MS may result in falsely low hepatic copper concentrations is clinically relevant, because low concentrations may result in a missed diagnosis of CuAH or lead to inappropriate treatment recommendations. It is possible that underestimation of hepatic copper concentration by spectroscopic-spectrometric methods has also blunted the perceived magnitude of the change in hepatic copper concentrations in dogs subsequent to modification of dietary copper recommendations for commercial dog food.

Rhodanine staining robustly identifies tissue copper and has previously been shown to provide a reliable semiquantitative estimate of hepatic copper concentration in Bedlington Terriers with CuAH.^23,24,25 With this stain, vivid orange-red copper aggregates contrast against a blue background of hematoxylin counterstain, enabling digital copper analysis. Rubeanic acid stain, another copper-specific stain, stains copper dark olive-green or black and requires a prolonged staining protocol, compared with rhodanine stain.⁸ Although 1 author⁸ has lauded the benefits of this stain, compared with rhodanine stain, the cytosolic black aggregates obtained with rubeanic acid stain may be mistaken for debris, tissue fixation artifacts, or bile accumulation and do not lend themselves to digital copper analysis. Unfortunately, the process of digital image analysis of rhodanine-stained hepatic sections for determination of copper concentration is not an automated procedure. Rather, with this technique, the operator must choose designated regions of interest and make sample-specific adjustments that optimize detection of hue and threshold color saturation. These settings avoid digital detection of lipofuscin and iron that otherwise might be mistaken for the orange color of stained copper. Before analysis of an individual sample, iterative assessment of settings is critical to verify that optimized adjustments only detect stained copper aggregates. Strong, intermediate, and weak positive and negative pixels in areas of interest are recorded with a proprietary algorithm.^f Data are then exported to a worksheet, and copper concentration is calculated by means of linear regression.⁹ During validation of this methodology, it became apparent that some hepatic copper concentrations obtained with AAS were erroneously low. These samples had nonhomogeneous copper distribution on inspection of rhodanine-stained sections and were excluded. Validation studies were completed with hepatic specimens that had a homogeneous distribution of copper and that underwent AAS multiple times.

Hepatic biopsy methods used for collection of specimens included in the present study reflected clinician preference. All 14-gauge needle specimens included ≥ 3 core samples with a total of ≥ 15 portal tracts, as required for accurate histologic assessment.¹⁸ However, we excluded 14-gauge needle specimens for assessment of intraindividual variation in hepatic copper distribution because these specimens are typically collected only from left-sided liver lobes. Broader sampling of liver lobes can be achieved with surgical or laparoscopic methods. In the present study, number of specimens submitted per dog was significantly higher for dogs that underwent laparoscopic cup biopsy than for dogs that underwent surgical or 14-gauge needle biopsy. Increasing the number of specimens examined would seem to increase the likelihood that digital image analysis of rhodanine-stained sections would accurately portray copper concentration.

In our experience, assessment of rhodanine-stained sections is optimized by simultaneously evaluating multiple sections on a single glass slide. Submission of multiple biopsy specimens in a single cassette or container makes it easier to process all specimens into a single paraffin block, which allows multiple sections to be placed on a single glass slide. However, it is important to acknowledge that assignment of a reliable qualitative copper accumulation score requires a standardized evaluation. Because of variations in tissue copper distribution and the qualitative nature of copper accumulation scoring, we did not anticipate an exact correlation between measured hepatic copper concentrations and assigned scores. However, the correlation was superior with concentrations obtained with digital image analysis than with concentrations obtained with AAS or ICP-MS.

In the present study, analysis of copper concentration in each section for dogs for which ≥ 4 rhodanine-stained sections were available for review revealed substantial variation among sections. Previously, the variation in mean hepatic copper concentration among liver lobes in healthy dogs was reported to range from 5% to 10%.⁸ We found comparable variation (median CV, 7.5%) in dogs with hepatic copper concentrations < 400 ppm (the upper reference limit) in the present study. However, significantly greater variation was encountered in dogs with pathological copper accumulation ≥ 1,000 ppm. An uneven hepatic copper distribution was present in 26.9% (139/516) of dogs in the present study, and histologic features contributing to uneven copper distribution were associated with significantly lower spectroscopic-spectrometric copper concentrations, compared with concentrations obtained with digital image analysis. Uneven copper distribution is also commonly reported in humans with pathological copper accumulation.^13,14,15 For dogs in the present study with pathological hepatic copper accumulation, the intraindividual variation in copper concentration obtained with digital image analysis ranged from 19.3% to 21.5%, which was similar to reports for human patients with Wilson disease, for which intraindividual variation ranges from 17% to 28%.¹⁶ Notably, this high intraindividual variation in measured copper concentrations raises concerns about using sequential needle biopsy specimens to assess changes in hepatic copper accumulation in dogs with CuAH over time.

Dogs in the present study that had necroinflammatory hepatic disease had significantly higher hepatic copper concentrations than did dogs without this condition, consistent with observations made by others.^2,3,g This finding in dogs likely reflects injury associated with the oxidative impact of copper rather than copper accumulation secondary to liver disease or cholestasis.^7,26,27 In fact, tissue samples from 3 dogs in the present study with cirrhosis lacked pathological hepatic copper accumulation.

Intraindividual variations in hepatic copper concentrations in the present study appeared to reflect the presence of hepatic fibrosis, regenerative nodules, and parenchymal extinction, as previously reported for human patients with Wilson disease.¹⁶ Areas with extensive bridging fibrosis typically had little copper, as these areas lack viable hepatic parenchyma. Because regenerative nodules are populated by comparatively young hepatocytes, these cells typically had low amounts of cytosolic copper, even in dogs with severe CuAH.^8,9 In dogs with tissue remodeling associated with CuAH, regenerative nodules are usually surrounded by hepatocytes and, to a lesser degree, macrophages that are heavily laden with copper.⁸ Consequently, examining areas of regenerative nodules can profoundly underestimate the degree of copper accumulation. Even in dogs with cirrhosis caused by CuAH, areas of parenchymal extinction and regenerative nodules may only have trace amounts of stainable copper. Our findings reinforce the idea that uneven hepatic copper distribution may be an important factor in the discordance between hepatic copper concentrations obtained by digital image analysis versus spectroscopic-spectrometric methods. We also discovered that uneven copper distribution may occur in some dogs without histologic features of necroinflammatory injury (data not shown).

Generally, hepatic copper concentrations obtained with AAS or ICP-MS were lower than paired concentrations obtained with digital image analysis of rhodanine-stained sections. In a few cases, however, the concentration obtained with the spectroscopic-spectrometric method was higher than the concentration obtained with digital image analysis. In our experience, we have occasionally encountered samples for which the hepatic copper concentration obtained with spectroscopic-spectrometric analysis was > 2,000 ppm, but rhodanine staining failed to substantiate pathological hepatic copper accumulation. In these unusual instances, quality control samples have verified that the rhodanine staining protocol was adequate, and rhodanine staining of additional sections has consistently confirmed our initial findings. Some of these discordant findings remain unexplained, although they might reflect mislabeled samples.

For many dogs in the present study, measurement of hepatic copper concentration was requested as a routine diagnostic procedure accompanying histologic evaluation of hepatic tissue. This likely reflected the fact that clinicians have been sensitized to the possibility of subclinical CuAH, which is reversible with appropriate treatment. For example, pathological copper accumulation has been serendipitously discovered in hepatic biopsy specimens collected from dogs undergoing prophylactic gastropexy, intestinal foreign body removal, lobectomy for treatment of hepatocellular carcinoma, cholecystectomy for treatment of a gallbladder mucocele, or portosystemic shunt attenuation.

The present study also found that hepatic copper concentrations were significantly higher in large-breed dogs (ie, dogs with a typical adult weight ≥ 20 kg). Because of the large number of Doberman Pinschers and Labrador Retrievers included in our study population, we were concerned that this finding could have been biased by the inclusion of these breeds, which are known to be predisposed to CuAH. However, even after dogs of these breeds were censored, large-breed dogs still had significantly higher hepatic copper concentrations than did smaller dogs. We also found that hepatic copper concentrations determined with digital image analysis were significantly higher in senior dogs (≥ 9 years old) than in younger dogs. These higher hepatic copper concentrations in larger and older dogs might reflect increased exposure to copper in commercial dog foods. However, further investigation with well-planned prospective studies are needed.

Although 2 older studies^25,27 suggested that histologic injury is not associated with hepatic copper concentrations < 2,000 ppm, as determined by AAS, our findings in the present study and in former studies^2,3 of Labrador Retrievers strongly contradict this conclusion. Indeed, we have frequently seen histologic evidence of copper-associated injury in dogs with hepatic copper concentrations obtained with digital image analysis as low as 600 ppm. In these dogs, recurring high hepatic transaminase activities were the impetus for hepatic biopsy, and the high enzyme activities resolved with copper chelation.

In conclusion, the present study confirmed that there may be clinically relevant differences between hepatic copper concentrations obtained with spectroscopic-spectrometric methods and concentrations obtained with digital image analysis of rhodanine-stained hepatic sections. Digital image analysis of rhodanine-stained sections is not widely available. Nevertheless, rhodanine-based qualitative copper accumulation scores can be used as a check on copper concentrations obtained by spectroscopic-spectrometric methods. Assignment of qualitative copper accumulation scores is facilitated by collecting biopsy specimens from 3 or 4 liver lobes and submitting all specimens in a single cassette or biopsy container so that specimens are embedded in a single paraffin block. This then allows multiple sections to be mounted and examined on a single slide, permitting comparison of copper distribution among biopsy sections and improving assignment of a qualitative copper accumulation score. Digital image analysis for determination of hepatic copper concentration should only be requested when clinically relevant copper accumulation is seen on rhodanine-stained slides. In our experience, this may include some specimens with a copper accumulation score of 2 but is typically more appropriate for specimens with scores ≥ 3. Digital image analysis of specimens without clinically relevant copper accumulation is unnecessary, because copper concentration is calculated with an algorithm that detects only stainable copper. The recently documented increase in hepatic copper concentration in dogs reported in the veterinary literature is largely based on spectroscopic-spectrometric analytic methods. Findings of the present study warrant concern that these reports represent a conservative appraisal of the severity of pathological copper accumulation in dogs.