Renal disease in dogs is most frequently evaluated noninvasively by serum biochemical analysis and urinalysis. This preliminary evaluation may be complemented by diagnostic imaging methods such as ultrasonography. These modalities have limitations in determining the underlying cause of renal disease, and assessment of cytologic and histologic samples can provide more detailed information about structural abnormalities and the etiopathogenesis of disease in the kidney. Histologic examination is considered the reference standard for diagnostic evaluation of solid tissues and for the kidney in particular1; however, the risk of hemorrhage is a consideration for this procedure. Hemostatic derangements in uremic people have been noted,2 and ultrasound-guided kidney biopsies in dogs and cats are more likely to cause major complications than liver biopsies, a finding possibly related to uremic thrombocytopathy, anemia, or concurrent systemic hypertension.3 The risk of complications is higher in thrombocytopenic than in nonthrombocytopenic dogs.3 One large-scale retrospective study4 found that kidney biopsy complications occurred in 38 of 283 (13%) dogs with predominantly nonneoplastic renal disease, with severe hemorrhage being the most common complication. The potential for kidney biopsy to affect organ function in animals with renal disease has not been fully evaluated, but this may be a concern.5
Cytologic evaluation is often used for the rapid microscopic screening of lesions, especially when there is a desire to minimize anesthetic exposure and mitigate the risk of biopsy-associated hemorrhage. Clinical reports6–11 of individual animals support the diagnostic utility of this procedure for a variety of neoplastic and nonneoplastic conditions. One review article12 suggested that cytologic examination of renal FNA samples may be useful for the diagnosis of lymphoma, carcinoma, metastatic neoplasia, cysts, and fungal infections in kidneys that are enlarged or have mass lesions, but that renal FNA samples are unlikely to contribute diagnostically useful information for most congenital or hereditary abnormalities, inflammatory conditions, and vascular lesions, particularly when the kidneys are small.12 However, large-scale studies systematically evaluating common patterns of usage, diagnostic recovery rate, diagnostic accuracy, and interpretation of cytologic samples in the context of imaging data are lacking.
The objective of the study reported here was to retrospectively evaluate the diagnostic value of cytologic examination of ultrasound-guided renal FNA samples from dogs, with histologic diagnosis or PARR results used as the reference standard. We additionally sought to characterize the ultrasonographic features of sampled kidneys and examine whether specific ultrasonographic features of the kidneys might be useful to aid cytologic interpretation.
Materials and Methods
Case selection and evaluation
Electronic medical records of the University of Minnesota Veterinary Medical Center were searched to identify all dogs that underwent cytologic examination of an ultrasound-guided renal FNA sample between January 12, 2005, and June 13, 2014. Cases were excluded if slides of cytologic preparations were not available for review in the slide archive or if only poorly cellular slides (< 1 nucleated cell/hpf) were available for review when the original cytology report described highly cellular slides.
Medical records were reviewed by 1 investigator (CAM). Information recorded included the kidney from which the sample was obtained (left vs right) and the breed, sex, and age of the dog. If available, histologic findings were reviewed and the histologic diagnosis was recorded. Animals for which the histologic reports indicated the findings were nondiagnostic were excluded from subjective comparisons between cytologic and histologic findings. For FNA samples in which cytologic examination raised suspicion for lymphoma and histologic examination was not performed, lymphoid clonality was assessed by PARR.
The study was reported according to the STARD (Standards for Reporting Diagnostic Accuracy Studies) guidelines.13,a The guidelines include a checklist of 25 items designed to help ensure internal and external validity of data while avoiding bias and improving the accuracy and completeness of reporting.
Review of cytologic specimens
All available archived Wright-Giemsa–stained cytologic specimens were reviewed by 2 investigators (a veterinary student [CAM] and a board-certified veterinary clinical pathologist [LCS]), and a consensus interpretation was determined. Both reviewers were blinded to the original diagnosis and any histologic findings, but were aware of signalment and diagnostic imaging findings. Each slide was categorized as adequate or inadequate for interpretation, depending on the quantity and cytologic detail of cellular elements. All slides deemed adequate for interpretation were evaluated for overall cellularity (nucleated cells), degree of blood contamination, and numbers of inflammatory, epithelial, round, and mesenchymal cells. For each category, samples were semiquantitatively scored according to the following scale: 0 = not seen, 1 = rare or mild, 2 = moderate, 3 = marked, and 4 = severe. In all slides deemed adequate for interpretation, anisocytosis and anisokaryosis were scored according to the described scale of 0 to 4; the nucleus-to-cytoplasm ratio was characterized as normal or increased, and the mean number and range of numbers of nuclei and nucleoli per cell were recorded. The presence or absence of background material (including proteinaceous fluid, necrotic debris, ultrasound gel, cytoplasmic fragments, extracellular matrix, and cholesterol crystals), hemorrhage (evidenced by erythrophagocytosis, iron pigment, or hematoidin), mitotic figures, infectious agents, and glomeruli was noted.
Collective findings across all adequate slides were used to assign a single retrospective cytologic diagnosis of inflammation, neoplasia, hyperplasia, necrosis, cyst, hemorrhage, or no abnormalities for each case. Evidence consistent with a cyst included the presence of proteinaceous fluid with few macrophages or inflammatory cells. Inflammation was categorized as suppurative or mixed on the basis of the predominant cell type: > 80% neutrophils was deemed suppurative, and < 80% neutrophils accompanied by other inflammatory cells was deemed mixed; the presence of any infectious agents was also noted. Within the category of neoplasia, cells were subclassified as round, epithelial, mesenchymal, neuroendocrine, or anaplastic. Neoplasia was categorized as benign or malignant according to routine diagnostic cytopathologic standards on the basis of the presence or absence of criteria of malignancy (anisocytosis, anisokaryosis, variable nucleus-to-cytoplasm ratio, cells with multiple nucleoli, or bizarre mitotic figures). The more criteria of malignancy are identified, the more likely a diagnosis of malignancy is to be rendered; however, there are tissue-specific and tumor-specific guidelines for some diagnoses that have not been established for kidneys because data are lacking.
Ultrasonographic examinations and image review
Although not specifically noted for every case, the typical protocol for renal aspiration at the University of Minnesota Veterinary Medical Center comprised full bilateral renal ultrasonography as part of a full abdominal scan, followed by aseptic preparation of the skin and transducer. Sedation was used at the discretion of the clinician; most commonly, butorphanol with or without midazolam was administered. A 22-gauge, 1.5-inch-long needle or spinal needle was inserted percutaneously through the path of least tissue resistance under real-time ultrasound guidance, followed by rapid forward and reverse thrusts of the needle through the region of interest. Negative pressure by use of a syringe with the needle in the region of interest was applied at the discretion of the radiologist. Color Doppler imaging was used to assess for hemorrhage before and after any subsequent aspiration if > 1 aspirate collection was performed or was used at the discretion of the radiologist for single aspirate procedures.
Archived still ultrasonographic images were reviewed by 2 authors (CAM and DAF [a board-certified veterinary radiologist]), and a consensus interpretation was determined. The time between FNA collection and diagnostic ultrasonography was noted for each case. Investigators were blinded to the cytologic findings and histologic diagnosis at the time of image review but were aware of patient signalment. Ultrasonographic images of the kidneys (evaluated individually) were assessed for size (subjectively judged as small, normal, or enlarged); presence of perirenal fluid; renal border shape (regular vs irregular and symmetric vs asymmetric); echogenicity (hyperechoic, isoechoic, or hypoechoic) of the renal cortex (relative to that of the liver and spleen) and medulla (relative to the cortex); characteristics of the renal pelvis (eg, presence of shadowing or complex fluid); corticomedullary interface characteristics (clear vs unclear); presence of renal pelvis dilation; presence and characteristics (hyperechoic, hypoechoic, or mixed echogenicity, compared with the renal cortex) of any masses present; and other abnormalities or notable findings. If 1 or both kidneys were obscured by masses on ultrasonographic images, a notation was made, and no further analysis was performed.
PARR
Archived renal aspirate slides from dogs with a cytologic diagnosis of lymphoma or round cell tumor were sent to the Colorado State University Clinical Immunology Laboratory for confirmatory testing by PARR; the median storage time for slides was 6 years (mean, 5 years; range, 0 to 9 years). Slides from dogs with a cytologic diagnosis other than lymphoma or round cell tumor were sent as negative controls with batches of sample slides. For the PARR, DNA was isolated from aspirates by use of commercial kitsb with a blood and body fluid protocol. Cells were scraped from 1 or 2 slides of cytologic specimens, and detection of rearranged immunoglobulin (IgH) and T-cell receptor genes was performed as previously described.14 All reactions included a positive DNA control (rhodopsin). A sample was considered to contain a clonal lymphoid population if the peak of the clonal product was > 3 times the height of the base polyclonal products. A sample was considered nondiagnostic if the positive control for DNA (rhodopsin) did not amplify or if a clonal PCR product with amplitude lower than that of the positive control product was detected.
Statistical analysis
A χ2 analysisc was used to determine whether there were significant differences in the number of cases for which specimens were collected from left versus right kidneys. The proportion of cytologic samples adequate for interpretation (diagnostic yield) was determined by dividing the number of specimens for which a cytologic diagnosis could be determined by the total number of specimens with representative slides that were reviewed. The proportion of adequate samples was also calculated for kidneys with ultrasonographic features that were identified in ≥ 4 dogs. A minimum of 4 dogs was chosen after 3 of the authors (CAM, LCS, and DAF) reviewed the tabulated data and deemed it appropriate to limit the influence and complexity of infrequent occurrences.
Cytologic and histologic or PARR findings were compared descriptively. Sensitivity, specificity, PPVs, and NPVs of cytologic examination were calculated. Because the number of cases with histologic or molecular confirmation of a diagnosis was small, calculations for diagnostic accuracy of cytologic examinations were restricted to the detection of nonneoplastic conditions overall, neoplasia of any type, and the most common tumor types diagnosed within the study (lymphoma and carcinoma). Histologic findings were considered definitive and used as the reference standard when available, and PARR results were considered definitive for a diagnosis of lymphoma and used as the reference standard when histologic examination was not performed. Diagnostic sensitivity was calculated as TP/(TP + FN), specificity was calculated as TN/(TN + FP), PPV was calculated as TP/(TP + FP), and NPV was calculated as TN/(TN + FN), where FN = false-negative results, FP = false-positive results, TN = true-negative results, and TP = true-positive results. For nonneoplastic conditions overall, considerations extended beyond the ability to distinguish between nonneoplastic and neoplastic processes; to be classified as a true-positive result, the cytologic diagnosis was required to reasonably agree with the reference standard diagnosis. Thus, cytologic identification of a nonneoplastic condition for which the reference standard diagnosis was either neoplasia or a substantially different nonneoplastic process was classified as a false-positive result. A true-negative result for this category also required that the neoplastic process identified cytologically was in general agreement with the reference standard diagnosis, and identification of a nonneoplastic process or of a substantially different tumor type by the reference standard method was cytologically classified as a false-negative result. Because a negative PARR result could exclude lymphoma but could not exclude other round cell tumors, and given the cytologic appearance of samples that would be assigned such a diagnosis, results for samples identified cytologically as representing lymphoma but that tested negative by PARR were classified as true-negative results.
Descriptive characterization of the ultrasonographic appearance of lymphoma, carcinoma, other neoplasia, and nonneoplastic lesions was performed by calculating the percentage of cases with a particular ultrasonographic feature evaluated in the study. Ultrasonographic features in kidneys with neoplastic versus nonneoplastic lesions were compared to determine whether there was a useful difference in ultrasonographic features (mass, pelvic dilation, perirenal fluid, diffuse increases or decreases in echogenicity) that could aid diagnosis when cytologic findings were equivocal.
Results
Study population
The medical records search identified 106 dogs that had 108 cytologic samples (2 dogs had renal aspirates collected twice from the same kidneys, once for primary diagnosis and a second time for evaluation of remission status). A flow diagram is provided (Supplementary Figure S1, available at avmajournals.avma.org/doi/suppl/10.2460/javma.252.10.1247). Six dogs were excluded from the study because of inability to retrieve representative cytologic specimens (5 had slides missing from the archive, and 1 had only nonrepresentative, poorly cellular slides available). The remaining 100 dogs and 102 samples were included in the study. One to 9 archived FNA smears were available for review for each case. Among 100 dogs, 51 breeds were represented, most commonly Labrador Retriever (n = 19), Golden Retriever (10), and German Shepherd Dog (6). Forty-three dogs were female (40 spayed and 3 sexually intact), and 57 were male (47 neutered and 10 sexually intact). The 2 dogs with > 1 sample evaluated were a spayed female and a neutered male. The median age of all dogs was 9 years (range, 7 months to 16 years).
Ultrasonographic findings and proportions of adequate cytologic specimens
All ultrasonographic examinations (n = 97) had still images available for review; these images represented 95 of the 100 dogs (the 2 dogs that underwent 2 renal FNAs had corresponding ultrasonographic examinations performed). The kidney that had an FNA sample collected was recorded for 80 cases. The left kidney was sampled significantly (P < 0.001) more frequently than the right (50 vs 22 dogs). Both kidneys were sampled in 8 cases. If there was a difference between the kidneys or if there was a difference in ultrasonographic renal appearance when 2 sampling dates were involved, these were considered individual occurrences (ie, cases); otherwise, bilateral sample collection was performed when the condition was considered bilateral, and these were treated as a single set of specimens. The number of ultrasonographic evaluations with each ultrasonographic characteristic of interest was variable, ranging from a minimum of 4 to 73. The most commonly identified features were a clear corticomedullary interface (n = 73), a mass lesion of any echogenicity (53), and hyperechogenicity of the cortex, relative to the spleen (44).
Of the 97 ultrasonographic evaluations available, 95 involved corresponding full renal ultrasonography performed the same day as the FNA. The remaining 2 specimens were collected under ultrasonographic guidance 7 and 15 days after the full diagnostic ultrasonography. The overall proportion of specimens with slides adequate for cytologic diagnosis (diagnostic yield) was 74 of 102 (73%). The diagnostic yield of cytologic slides for the 97 samples that had corresponding ultrasonographic images available was stratified by renal ultrasonographic characteristics (Table 1). The highest diagnostic yield was found in kidneys with unclear corticomedullary distinction (18/20 [90%]), pelvic dilation (18/20 [90%]), and an infiltrative or nodular appearance (6/7 [86%]).
Diagnostic yield of cytologic specimens by ultrasonographic characteristics of the same kidney or kidneys in 97 examinations of 95 dogs that had renal FNA and corresponding ultrasonographic examinations performed.
| Ultrasonographic characteristic | No. of specimens from kidneys with an ultrasonographic characteristic | No. of specimens adequate for diagnosis (diagnostic yield [%]) |
|---|---|---|
| Clear corticomedullary interface | 73 | 49 (67%) |
| Any mass or masses (any echogenicity)* | 53 | 42 (79%) |
| Hyperechoic kidney† (normal-sized or enlarged) | 44 | 31 (71%) |
| Unilateral mass* (any echogenicity) | 35 | 28 (80%) |
| Hypoechoic kidney‡ (normal-sized or enlarged) | 21 | 16 (76%) |
| Pelvic dilation | 20 | 18 (90%) |
| Unclear corticomedullary interface | 20 | 18 (90%) |
| Bilateral masses (any echogenicity)* | 18 | 14 (78%) |
| Unilateral mass (isoechoic or mixed echogenicity)* | 18 | 14 (78%) |
| Unilateral hypoechoic mass* | 13 | 11 (85%) |
| Bilateral hypoechoic masses* | 10 | 8 (80%) |
| Infiltrative or nodular appearance (normal-sized or enlarged) | 7 | 6 (86%) |
| Bilateral masses (isoechoic or mixed echogenicity)* | 7 | 5 (71%) |
| Isoechoic or hyperechoic medulla* | 5 | 3 (60%) |
| Unilateral hyperechoic mass* | 4 | 3 (75%) |
Findings are reported in order of decreasing frequency of observation of the imaging feature.
Echogenicity relative to the renal cortex.
Isoechoic or hyperechoic to the spleen.
Hypoechoic to the liver.
Cytologic findings
Cytologic specimens were typically of moderate (26/74 [35%]) or severe (24/74 [32%]) cellularity. Specimens that were markedly (12/74 [16%]) or minimally (9/74 [12%]) cellular were less common. Acellular diagnostic specimens (eg, samples from cysts or necrotic tissue) were the least common (3/74 [4%]). Almost all samples had some amount of blood in the background that was considered to represent peripheral blood contamination, but mild to moderate hemorrhage (presence of erythrophagocytosis, hemosiderin, or hematoidin) was observed in 11 of 74 (15%) cases. Background material was uncommonly identified but included necrotic debris (9/74 [12%]), proteinaceous fluid (8/74 [11%]), ultrasound gel (6/74 [8%]), and cholesterol crystals (1/74 [1%]). Renal tubular epithelial cells were noted in 48 of 74 (65%) specimens and typically classified as mild (19/74 [26%]) to moderate (25/74 [34%]) in number, with only 4 of 74 (5%) specimens containing marked to severe numbers of tubular epithelial cells. No glomeruli were observed. Cocci were identified in 1 case, and rods were identified in another.
Of the 74 specimens considered adequate for interpretation, the authors cytologically diagnosed carcinoma in 17 (specimens from 17 dogs); these occasionally had other abnormalities present, including necrosis (n = 3), inflammation (3), hemorrhage (3), and renal tubular hyperplasia (1) sufficiently prominent to be included in the morphological diagnosis. Nineteen specimens were from 17 dogs with a cytologic diagnosis of lymphoma; one of these diagnoses was associated with renal tubular hyperplasia. Other diagnoses of malignancy included sarcoma (n = 2), anaplastic malignancy (2), and round cell tumor (1). The remaining specimens were from dogs with other primary diagnoses, including renal tubular hyperplasia (n = 16, with 1 including necrosis and inflammation), inflammation (3), necrosis (5), hemorrhage (1), renal cyst (1), or with no cytologic abnormalities evident (8).
Diagnostic accuracy of cytologic examination
Twenty-eight of 102 cases had histologic examination reports available for review. Two of the 28 histologic specimens were described as nondiagnostic; these also had cytologic specimens deemed inadequate for review. Of the remaining 26 cases (involving 26 dogs) with definitive histologic diagnoses, 7 had cytologic specimens deemed inadequate for interpretation, leaving 19 cases for evaluation of correspondence between diagnoses by cytologic and histologic methods (Table 2). Specimens evaluated histologically were collected between 48 hours and 3 years before (n = 1) or after (18) the time of FNA, with most (15/19) collected ≤ 2 weeks after the FNA.
Comparison of retrospective cytologic diagnosis for renal FNA samples versus diagnosis by a reference standard method (histologic evaluation [n = 19] or PARR [7]) for 26 dogs that had data available.
| Diagnosis (No. of cases) | ||
|---|---|---|
| Cytologic examination | Reference standard result | Histologic sample collection interval (d)* |
| Lymphoma (10)† | Lymphoma by PARR (5) | 0 (0–1) |
| Negative by PARR (2) | ||
| Lymphoma by histologic examination (3) | ||
| Carcinoma (6) | Renal carcinoma (4) | 6 (1–80) |
| Acute tubular necrosis with regeneration (1) | ||
| Fibrinosuppurative nephritis (1) | ||
| Sarcoma (1) | Hemangiosarcoma (1) | 172 |
| Anaplastic malignancy (1) | Round cell tumor‡ (1) | 5 |
| Hyperplasia (4) | Histiocytic sarcoma (1) | 6 (1–16) |
| Pleomorphic transitional cell carcinoma (1) | ||
| High-grade sarcoma (1) | ||
| Tubular degeneration (1) | ||
| Suppurative inflammation (1) | Marked focal chronic fibrosis of a large wedge-shaped region of cortex consistent with resolution of previous renal abscess; concurrent lymphoplasmacytic nephritis (1) | 1,084 |
| Necrosis (2) | Carcinoma (1) | 3 (1–5) |
| Renal cyst, amyloidosis, and lymphoplasmacytic inflammation (1) | ||
| Normal (1) | Moderate lymphoplasmacytic infiltrate (1) | 1 |
Histologic examination was performed on renal samples obtained at necropsy (n = 10), renal biopsy samples (6), or biopsy samples from another site but considered representative of the renal lesion in dogs with disseminated disease (3; an ocular lesion [1 dog with renal carcinoma], a splenic lesion [1 dog with hemangiosarcoma], and a perianal mass [1 dog with round cell tumor]). Cytologic examination preceded histologic examination or PARR in all cases except one (a dog with splenic hemangiosarcoma).
Time is reported as number of days or as median (range) for multiple dogs and represents the interval between FNA and histologic sample collection.
Nine of 10 dogs that had a cytologic diagnosis of lymphoma had sufficient DNA for PARR; 3 had histologic confirmation of lymphoma (for 2 dogs with results for both methods, histologic examination was used as the reference standard; 1 of these had negative PARR results, and 1 had PARR results indicative of T-cell lymphoma). Of the remaining 7 dogs tested by PARR, results for 5 were indicative of B-cell (n = 2) or T-cell (n = 3) lymphoma, and results for the remaining 2 were negative.
Tumor could not be subclassified.
Eighteen cases for which a cytologic diagnosis of lymphoma or round cell tumor was made had specimens submitted for evaluation by PARR. Nine cases had insufficient DNA for PARR and were excluded from further analysis unless a histologic diagnosis was obtained (n = 1). Six of the 9 cases with sufficient DNA were diagnosed as lymphoma by PARR, including 1 that had histologic confirmation. Of the 3 cases tested negative by PARR, 1 was histologically diagnosed as lymphoma, suggesting a false-negative PARR result. Of 7 negative control specimens containing epithelial tissue (identified by histologic [n = 3] or cytologic [4] methods), 2 had insufficient DNA for PARR and 4 tested negative. In the remaining sample, a clonal T-cell population was identified. This suggested a false-positive result; however, no histologic or long-term follow-up information was available to confirm that the finding was erroneous. The diagnostic accuracy of cytologic findings for detection of lymphoma, carcinoma, neoplasia overall, and nonneoplastic conditions overall was summarized (Table 3).
Sensitivity, specificity, PPV, and NPV of cytologic examination for the detection of neoplastic and nonneoplastic conditions in 26 renal FNA samples from 26 dogs.
| Sensitivity | Specificity | PPV | NPV | |||||
|---|---|---|---|---|---|---|---|---|
| Disease | TP/(TP + FN) | % | TN/(TN + FP) | % | TP/(TP + FP) | % | TN/(TN + FN) | % |
| Neoplasia (overall) | 14/18 | 78 | 4/8 | 50 | 14/18 | 78 | 4/8 | 50 |
| Lymphoma | 8/8 | 100 | 16/18 | 89 | 8/10 | 80 | 16/16 | 100 |
| Carcinoma | 4/6 | 67 | 18/20 | 90 | 4/6 | 67 | 18/20 | 90 |
| Nonneoplastic condition (overall) | 2/4 | 50 | 17/22 | 77 | 2/7 | 29 | 17/19 | 89 |
Histologic results were used as the reference standard when available (n = 19); for 7 dogs with a cytologic diagnosis of lymphoma that had no histologic sample available, PARR results were used as the reference standard.
For analysis of nonneoplastic conditions overall, requirements for a true-positive result included reasonable agreement between the cytologic and reference standard diagnoses as well as identification of a nonneoplastic process (ie, cytologic identification of a nonneoplastic condition for which the reference standard diagnosis was either neoplasia or a substantially different nonneoplastic process was classified as a false-positive result). Similarly, a true-negative result required that the neoplastic process identified cytologically was in general agreement with the reference standard diagnosis; thus, identification of a nonneoplastic process or of a substantially different tumor type by the reference method resulted in classification of the cytologic finding as false-negative.
FN = False-negative. FP = False-positive. TN = True-negative. TP = True-positive.
See Table 2 for remainder of key.
Cytologic examination resulted in incorrect diagnosis of carcinoma in 2 specimens, one identified as representing acute tubular necrosis and the other as fibrinosuppurative nephritis on histologic evaluation; this was possibly related to the presence of epithelial hyperplasia. Two false-negative cytologic diagnoses of hyperplasia and necrosis were also made in histologically confirmed cases of carcinoma. Although the numbers were insufficient for statistical analysis, the range of values for the cytologic characteristics of the samples with true-positive (n = 4), false-positive (2), and false-negative (1) diagnoses for carcinoma was summarized descriptively (Table 4). Although there were 2 false-negative cytologic results, the second sample consisted only of necrotic debris, so no cellular characteristics could be defined. Subjectively, specimens with an accurate cytologic diagnosis of carcinoma had high cellularity, and some inflammation was observed in all cases. Otherwise, there was overlap in the range of values reported for cytologic criteria of malignancy in samples collected from kidneys with and without histologically confirmed carcinoma. Examples of cytologic specimens from dogs with histologic diagnoses of renal epithelial hyperplasia and carcinoma are shown (Figure 1).
Photomicrographs of cytologic preparations depicting the overlap in cytologic features of renal epithelial hyperplasia and carcinoma. A—Normal to mild hyperplastic renal tubular epithelium. B and C—Atypical renal tubular epithelial cells in specimens that had false-positive cytologic diagnoses of carcinoma. D through F—Neoplastic renal epithelial cells from 3 dogs with true-positive cytologic diagnoses of carcinoma. Wright-Giemsa stain; bar = 20 μm.
Citation: Journal of the American Veterinary Medical Association 252, 10; 10.2460/javma.252.10.1247
Photomicrographs of cytologic preparations depicting the overlap in cytologic features of renal epithelial hyperplasia and carcinoma. A—Normal to mild hyperplastic renal tubular epithelium. B and C—Atypical renal tubular epithelial cells in specimens that had false-positive cytologic diagnoses of carcinoma. D through F—Neoplastic renal epithelial cells from 3 dogs with true-positive cytologic diagnoses of carcinoma. Wright-Giemsa stain; bar = 20 μm.
Citation: Journal of the American Veterinary Medical Association 252, 10; 10.2460/javma.252.10.1247
Photomicrographs of cytologic preparations depicting the overlap in cytologic features of renal epithelial hyperplasia and carcinoma. A—Normal to mild hyperplastic renal tubular epithelium. B and C—Atypical renal tubular epithelial cells in specimens that had false-positive cytologic diagnoses of carcinoma. D through F—Neoplastic renal epithelial cells from 3 dogs with true-positive cytologic diagnoses of carcinoma. Wright-Giemsa stain; bar = 20 μm.
Citation: Journal of the American Veterinary Medical Association 252, 10; 10.2460/javma.252.10.1247
Cytologic features of epithelial cells in 7 of 8 FNA samples from 8 dogs that had true-positive, false-positive, and false-negative cytologic diagnoses of carcinoma, compared with histologic findings (the reference standard).
| Result | No. of specimens | Cellularity | Inflammation | Anisocytosis | Anisokaryosis | N:C | Nuclei (No. per cell) | Multiple nucleoli (No. of samples) |
|---|---|---|---|---|---|---|---|---|
| True-positive | 4 | 3–4 | 1–3 | 1–3 | 1–3 | Normal to high | 1 | Not present (3) Present (1) |
| False-positive | 2 | 2 | 0–3 | 2–3 | 2–3 | Normal to high | 1 | Present |
| False-negative | 1* | 3 | 0 | 2 | 2 | High | 1 | Not present |
Cellularity, inflammation, anisocytosis, and anisokaryosis were scored as follows: 0 = not present, 1 = minimal, 2 = moderate, 3 = marked, and 4 = severe. Ranges are reported for multiple samples.
Two specimens had false-negative results; only 1 contained cells that could be characterized.
N:C = Nucleus-to-cytoplasm ratio.
Although not included in calculations of diagnostic accuracy, a urine culture result was positive for Staphylococcus intermedius for the specimen in which cocci were identified cytologically, and culture of the aspirated material had positive results for Escherichia coli for the specimen in which bacterial rods were identified cytologically.
Renal ultrasonographic characteristics by histologic diagnosis
To assess whether the presence or absence of certain ultrasonographic features might be used to inform the final disease diagnosis, we examined the frequency of ultrasonographic findings in 16 dogs with diagnoses of lymphoma (n = 7), carcinoma (5), or other neoplasia (4) confirmed by the reference standard methods (Table 5). Kidneys of all dogs found to have nonneoplastic lesions by histologic examination (n = 5) were enlarged or of normal size and hyperechoic; 4 had a clear corticomedullary interface. Diffuse renal enlargement with pelvic dilation, pelvic dilation alone, and unclear corticomedullary interface were each identified in a single case; no other ultrasonographic features were observed in these cases. All histologically confirmed carcinomas that were evaluated ultrasonographically (n = 5) had masses identified, whereas no masses were found for cases in which renal lesions were nonneoplastic. Thus, the 2 false-negative cytologic diagnoses of carcinoma were for specimens from kidneys with mass lesions, and the 2 false-positive cytologic diagnoses of carcinoma were for specimens from kidneys that lacked masses. A unilateral mass was identified in 5 of 5 confirmed cases of carcinoma, versus 2 of 7 confirmed cases of lymphoma and 2 of 4 cases in which other types of neoplasia were diagnosed. Cortical, medullary, or overall renal echogenicity appeared to be of little value in distinguishing among the various neoplastic processes in the absence of focal lesions.
Ultrasonographic findings for 16 confirmed cases of lymphoma, carcinoma, or other neoplasia in 16 dogs.
| Finding | Lymphoma (n = 7) | Carcinoma (n = 5) | Other neoplasia (n = 4) |
|---|---|---|---|
| Unilateral mass (any echogenicity) | 2 | 5 | 2 |
| Bilateral masses (any echogenicity) | 3 | 0 | 1 |
| ≥ 1 mass (any echogenicity) | 5 | 5 | 3 |
| Unilateral hyperechoic mass | 0 | 2 | 0 |
| Unilateral isoechoic or mixed echogenicity mass | 1 | 3 | 2 |
| Unilateral hypoechoic mass | 1 | 0 | 0 |
| Bilateral hypoechoic masses | 3 | 0 | 1 |
| Hypoechoic kidney (normal size or enlarged) | 2 | 1 | 1 |
| Hyperechoic kidney (normal size or enlarged) | 3 | 1 | 1 |
| Infiltrative or nodular appearance of kidney (normal size or enlarged) | 1 | 0 | 0 |
| Renal pelvis dilation | 1 | 2 | 1 |
| Clear corticomedullary interface | 5 | 5 | 2 |
| Unclear corticomedullary interface | 1 | 0 | 2 |
Only features that were observed are listed. The category of other neoplasia included hemangiosarcoma, round cell tumor, pleomorphic transitional cell tumor, and high-grade sarcoma. Masses were deemed hypoechoic, hyperechoic, or isoechoic relative to the renal cortex. Kidneys were deemed hypoechoic if less echogenic than the liver and hyperechoic if isoechoic or hyperechoic relative to the spleen. Diagnoses for 5 cases of lymphoma were confirmed by PARR alone; all remaining diagnoses were confirmed histologically.
Discussion
In the present study, the overall proportion of cytologic preparations of renal FNA samples deemed adequate for interpretation was considered good at 74 of 102 (73%). This was consistent with frequencies reported for cytologic preparations from other tissues, including 243 of 292 (83%) for nonmammary cutaneous and subcutaneous lesions in dogs and cats,15 928 of 1,274 (73%) for lymph node specimens in dogs,16 and 69 of 95 (73%) for pancreatic specimens in dogs.17 One small study18 of the ultrasonographic characteristics of renal lymphoma in dogs found that 7 of 9 aspirates were adequate for interpretation on the first attempt, similar to our findings in a larger and more heterogeneous group of dogs.
To date, no evidence-based guidelines have been published to provide diagnostic imaging–based indications for renal FNA and cytologic evaluation of canine kidneys. Not surprisingly, in the absence of such guidelines, we found that this procedure was performed at our institution in dogs with a broad spectrum of ultrasonographic findings. Our results supported that the proportion of adequate cytologic specimens varies greatly across this spectrum. Specifically, our results suggested that ultrasonographic identification of focal or multifocal lesions was associated with the highest diagnostic yields, whereas alterations in corticomedullary clarity or relative medullary echogenicity resulted in the lowest yields, although statistical comparisons were not made.
One previous report12 suggested that one of the limitations of renal cytologic examination is a high likelihood of samples with low cellularity. In the present study, 62 of 74 (84%) specimens were of moderate to severe cellularity. The study was retrospective, and prescreening submission bias and different levels of experience among individuals who collected and prepared the specimens could not be excluded. However, the cellularity of renal FNA samples was considered to be sufficient for interpretation by board-certified veterinary clinical pathologists in most of the cases and slides that were included in our study.
Although previous studies7,19 of renal lymphoma in dogs have relied on cytologic findings for a definitive diagnosis, we sought to maximize our ability to confirm cytologic diagnoses by performing PARR on samples with a cytologic diagnosis of lymphoma. Notably, a high percentage of tested specimens or negative control samples (11/25 [44%]) in our study was considered nondiagnostic owing to insufficient retrieval of DNA. Although we are unaware of studies on this subject in veterinary diagnostic medicine, a number of preanalytical variables have been described that can affect the quality and yield of DNA extracted from human cytologic specimens, including overall sample cellularity, fixative and stain types, and slide type.20,21 The specimens submitted for PARR in the present study were air-dried and stained with Wright-Giemsa stain; according to comparable human studies, these variables might have negatively impacted DNA yield. Although storage time has been reported to impact DNA yield, the mean age of slides with insufficient DNA in our study was 6 years and the 2 oldest slides generated PARR results. Therefore, we did not believe that sample age was an important factor in our study. Moreover, all slides submitted for PARR were prescreened by the authors, and we did not believe that low-sample cellularity was a contributing factor. Finally, although some tissue-specific variables could potentially affect the DNA yield from renal FNA samples, the authors do not believe this factor would substantially impact the results.
The imaging characteristics of renal lymphoma in a case series report18 of 10 dogs with renal lymphoma identified pyelectasia in all dogs, loss of corticomedullary distinction in 9, enlarged kidneys in 8, and renal deformity and hypoechoic lesions in 6. Bilateral lesions were identified in 9 of those dogs, and 3 dogs were described as having only minor ultrasonographic abnormalities. In the present study, a clear corticomedullary interface was frequently present with any of the proven neoplasms, but the presence of bilateral masses (regardless of echogenicity) was almost twice as common in kidneys with lymphoma as in those with other neoplastic processes.
One multi-institutional study22 of 82 dogs with primary renal neoplasia identified carcinoma in 49, sarcoma in 28, and nephroblastoma in 5. Although primary renal neoplasia is rare, carcinoma was the most common type in that study.22 Although our data set was very limited, we were able to accurately diagnose carcinoma cytologically in 4 cases. There were 2 false-negative and 2 false-positive cytologic diagnoses of carcinoma as well. The false-negative results could have been attributable to sampling bias or a tendency for cytologists to be conservative in cytologic interpretation to avoid overdiagnosis of cancer. Thus, specificity could be expected to exceed sensitivity for cytologic diagnosis of renal neoplasia, as has been demonstrated in other tissues, although the present study only revealed this regarding carcinoma.23–25 The 2 false-positive diagnoses of carcinoma were concerning, considering that clinical action can be instigated on the basis of a cytologic diagnosis, often in combination with imaging data, when more definitive diagnostic tests are deemed too invasive or costly. The grading of key cytologic features of malignancy, such as anisocytosis, anisokaryosis, and the nucleus-to-cytoplasm ratio, and the presence of multiple nucleoli overlapped between neoplastic and nonneoplastic lesions in the present study. This suggested that the diagnosis of renal carcinoma by cytologic examination alone can be challenging. The histopathologic features of renal carcinoma can be quite variable in regard to cellular pleomorphism, with only mitotic index being identified as prognostic in 1 recent study,26 suggesting that marked cytologic atypia does not characterize all variants. This suggests it may be difficult to distinguish carcinomas from adenomas on the basis of cytologic findings alone. Renal adenomas have been described as subclinical round compressive lesions comprised of minimally pleomorphic epithelial cells with rare or absent mitotic figures and no accompanying hemorrhage or necrosis.27 A review12 of the subject acknowledged the difficulty in distinguishing benign from malignant renal neoplasms because of the overlap in cytologic appearance and indicated that larger lesions are sometimes presumed to be malignant. The presence of atypical epithelial tissue on histologic examination has been briefly described as associated with canine renal dysplasia, but specific cytologic characteristics were not reported28 and have received little attention in the literature elsewhere.
Our results suggest that the combination of cytologic and ultrasonographic characteristics provides complementary information when assessing focal lesions and their association with neoplasia and for deciding which dogs need histologic sampling (focal lesions without confident cytologic diagnosis) versus clinical and imaging follow-up. Furthermore, we also observed that renal ultrasonography, when combined with cytologic evaluation of FNA specimens, may aid in distinguishing renal carcinomas from reactive epithelial hyperplasia. In our data, the presence of focal or multifocal architectural disruptions fostered the yield of ultrasound-guided cytologic sampling as well as added context to the likelihood of neoplasia. Carcinoma was the most frequently proven neoplastic diagnosis, particularly when the lesion was unilateral and hyperechoic. Unfortunately, we did not have enough comparable inflammatory focal lesions to allow comparison between the combination of ultrasonographic findings and cytologic yield when differentiating focal or multifocal inflammation from neoplasia.
Evaluation of diagnostic accuracy of cytologic examination in retrospective studies can be influenced by selection bias, as not all dogs that have cytologic evaluation of a lesion also have histologic examination performed.23 In the present study, 10 of 19 histologic diagnoses were determined at necropsy and 3 were determined by biopsy of tissues other than the kidneys in dogs with disseminated disease; incisional or excisional renal biopsies were performed in only 6 cases. Ideally, the cytologic diagnoses of renal lesions should be evaluated prospectively in larger groups of dogs undergoing simultaneous histologic evaluation, preferably with similar numbers of dogs evaluated for suspected neoplastic and nonneoplastic lesions. Nonetheless, this study represented one of the largest to date in which cytologic evaluation of renal FNA samples and diagnostic imaging in dogs were performed to assess the frequency with which cytologic preparations were deemed adequate for interpretation, to identify common cytologic and ultrasonographic features in samples from dogs with neoplastic and nonneoplastic renal lesions, and to evaluate the diagnostic accuracy of cytologic evaluation for such patients.
Acknowledgments
Supported in part by the Summer Scholars Program with internal funds from the College of Veterinary Medicine, University of Minnesota, Saint Paul.
Presented in poster form at the Combined Annual Meeting of the American College of Veterinary Pathologists, the American Society for Veterinary Clinical Pathology, and the Society for Toxicologic Pathology, Minneapolis, Oct 2015.
The authors declare that there were no conflicts of interest.
ABBREVIATIONS
| FNA | Fine-needle aspirate |
| NPV | Negative predictive value |
| PARR | PCR assay for antigen receptor rearrangement |
| PPV | Positive predictive value |
Footnotes
STARD flow diagram [database online]. 2003. Available at: www.stard-statement.org. Accessed Aug 15, 2013.
QIAmp DNA mini kit QIAGEN Sciences, Germantown, Md.
MedCalc for Windows, version 16.1, MedCalc Software, Ostend, Belgium.
References
1. Manashirova M, Pressler BM, Gelb HR, et al. Pilot evaluation of a vacuum-assisted biopsy instrument for percutaneous renal biopsy in dogs. J Am Anim Hosp Assoc 2011;47:391–398.
2. Lutz J, Menke J, Sollinger D, et al. Haemostasis in chronic kidney disease. Nephrol Dial Transplant 2014;29:29–40.
3. Bigge LA, Brown DJ, Pennick DG. Correlation between coagulation profile findings and bleeding complications after ultrasound-guided biopsies: 434 cases (1993–1996). J Am Anim Hosp Assoc 2001;37:228–233.
4. Zatelli A, Bonfanti U, Santilli R, et al. Echo-assisted percutaneous renal biopsy in dogs. A retrospective study of 229 cases. Vet J 2003;166:257–264.
5. Vaden SL, Levine JF, Lees GE, et al. Renal biopsy: a retrospective study of methods and complications in 283 dogs and 65 cats. J Vet Intern Med 2005;19:794–801.
6. Battaglia L, Petterino C, Castagnaro M. Hypoglycaemia as a paraneoplastic syndrome associated with renal adenocarcinoma in a dog. Vet Res Commun 2005;29:671–675.
7. Collicutt NB, Gamer BC, Brown CA, et al. What is your diagnosis? Renal mass in a dog. Vet Clin Pathol 2013;42:389–390.
8. Hess RS, Ilan I. Renal abscess in a dog with transient diabetes mellitus. J Small Anim Pract 2003;44:13–16.
9. Merrick CH, Schleis SE, Smith AN, et al. Hypercalcemia of malignancy associated with renal cell carcinoma in a dog. J Am Anim Hosp Assoc 2013;49:385–388.
10. Michael HT, Sharkey LC, Kovi RC, et al. Pathology in practice. Renal nephroblastoma in a young dog. J Am Vet Med Assoc 2013;242:471–473.
11. Petterino C, Luzio E, Baracchini L, et al. Paraneoplastic leukocytosis in a dog with a renal carcinoma. Vet Clin Pathol 2011;40:89–94.
12. Borjesson DL. Renal cytology. Vet Clin North Am Small Anim Pract 2003;33:119–134.
13. Bossuyt PM, Reitsma JB, Bruns DE, et al. Towards complete and accurate reporting of studies of diagnostic accuracy: the STARD initiative. Clin Chem 2003;49:1–6.
14. Burnett RC, Vernau W, Modiano JF, et al. Diagnosis of canine lymphoid neoplasia using clonal rearrangements of antigen receptor genes. Vet Pathol 2003;40:32–41.
15. Ghisleni G, Roccabianca P, Ceruti R, et al. Correlation between fine-needle aspiration cytology and histopathology in the evaluation of cutaneous and subcutaneous masses from dogs and cats. Vet Clin Pathol 2006;35:24–30.
16. Amores-Fuster I, Cripps P, Graham P, et al. The diagnostic utility of lymph node cytology samples in dogs and cats. J Small Anim Pract 2015;56:125–129.
17. Cordner AP, Sharkey LC, Armstrong PJ, et al. Cytologic findings and diagnostic yield in 92 dogs undergoing fine-need aspiration of the pancreas. J Vet Diagn Invest 2015;27:236–240.
18. Taylor AJ, Lara-Garcia A, Benigni L. Ultrasonographic characteristics of canine renal lymphoma. Vet Radiol Ultrasound 2014;55:441–446.
19. Avery A. Molecular diagnostics of hematologic malignancies. Top Companion Anim Med 2009;24:144–150.
20. Roy-Chowdhuri S, Chow CW, Kane MK, et al. Optimizing the DNA yield for molecular analysis from cytologic preparations. Cancer Cytopathol 2016;124:254–260.
21. Dejmek A, Zendehrokh N, Tomaszewska M, et al. Preparation of DNA from cytological material: effects of fixation, staining, and mounting medium on DNA yield and quality. Cancer Cytopathol 2013;121:344–353.
22. Bryan JN, Henry CJ, Turnquist SE, et al. Primary renal neoplasia of dogs. J Vet Intern Med 2006;20:1155–1160.
23. Bahr KL, Sharkey LC, Murakami T, et al. Accuracy of US-guided FNA of focal liver lesions in dogs: 140 cases (2005–2008). J Am Anim Hosp Assoc 2013;49:190–196.
24. Sharkey LC, Dial SM, Matz ME. Maximizing the diagnostic value of cytology in small animal practice. Vet Clin North Am Small Anim Pract 2007;37:351–372.
25. Simon D, Schoenrock D, Nolte I, et al. Cytologic examination of fine-needle aspirates from mammary gland tumors in the dog: diagnostic accuracy with comparison to histopathology and association with postoperative outcome. Vet Clin Pathol 2009;38:521–528.
26. Edmondson EF, Hess AM, Powers BE. Prognostic significance of histologic features in canine renal cell carcinomas: 70 nephrectomies. Vet Pathol 2015;52:260–268.
27. Baskin GB, DePaoli A. Primary renal neoplasms of the dog. Vet Pathol 1977;14:591–605.
28. Picut CA, Lewis RM. Microscopic features of canine renal dysplasia. Vet Pathol 1987;24:156–163.
