Study design
From 133 samples of altered and unaltered parts of 66 canine spleens FNP and imprint smears were collected and the residual tissue was forwarded to the histopathological examination. FNP and imprint smears were cytologically assessed regarding sample
quality and diagnosis. Cytologic accuracy was determined based on total and partial agreement or disagreement when compared to results of the histopathologic diagno-sis.
Additionally, possible influences of the sample collection procedure and sample quality on the diagnostic accuracy were examined by comparing the percentages of agree-ment for FNP and imprints and also samples of moderate to high and poor quality.
Sample material
133 splenic specimens from 66 canine spleens (1–3 samples from each spleen) en-tered the study between October 2012 and October 2015 for comparative cytological and histopathological examination. 69/133 samples were taken from splenic malignant or benign tumors or other alterations, 48 samples originated from macroscopically un-altered residual parts of those spleens and 16 samples originated from 9 splenomeg-alies (of 7 organs 2 samples were collected). The sample size was one to eight cubic centimetres. Nearly all splenic tissue resulted from surgically removed spleens.
The vast majority of the spleens originated from splenectomy due to medical indica-tions including tumor (n = 54), torsion of the spleen or torsio ventriculi with splenomeg-aly and splenic alteration (n = 4) or other causes of splenomegsplenomeg-aly (n = 5). All splenec-tomies were performed in the Small Animal Clinic, University of Veterinary Medicine Hannover, except three cases which were splenectomized at the Clinic for Small Ani-mals, Isernhagen, before samples were collected by the first author (S.C.). The re-maining three spleens all of them containing at least one nodular lesion were collected within two hours after euthanasia to prepare samples. Two of these animals had to be euthanized with neoplastic disease affecting the spleen, one dog had prostatitis. All
samples were included, when adequate sample processing could be guaranteed and otherwise have not been pre-selected in any way.
Preparation of cytological specimens
Fine-needle puncture was carried out with a 5 ml syringe attached to a 22 Gauge nee-dle by the first author or – in some cases – by laboratory staff members as well as trainees. As the spleen is a parenchymal organ, the puncture was done without aspi-ration to avoid blood contamination18. Samples were prepared using the squash tech-nique following the standard procedure22. Impression smears of the splenic samples were prepared from freshly prepared tissue cut surfaces after blood and fluid had been removed by dabbing several times on absorbent material22. After air drying, the sam-ples were stained using the Pappenheim stain (May Gruenwald and Giemsa stain).
Cytological examination
Cytological examination was interpreted blinded, i. e. without clinical information in-cluding localization (e. g. local lesion vs morphologic unaltered splenic tissue), tech-nique (FNP or impression smear) and especially without knowledge of the histopatho-logical diagnoses by one of the coauthors (R.M.), who is the head of the clinical pa-thology laboratory of the University Hospital for Small Animals, Horses, and exotic pets and has a national degree for clinical pathology and is an Associate member of the ECVCP.
Histopathological reference examination
Histopathological specimens were collected from the same piece of spleen and were treated with standard methods. In addition, in most cases the remaining splenic tissue was examined routinely beyond this study. The tissue was fixed in 10 % formalin and, after embedding in paraffin 3–4 µm thin sections were cut and stained with hematoxylin and eosin (HE). Immunohistochemistry was executed in two cases to verify final diag-nosis, because diagnosis varied from routine histopathological diagnosis of the remain-ing spleen. Histopathologic probes were evaluated by one of the coauthors (M. H-T.) who is an experienced boarded pathologist of the Department of Pathology, University for Veterinary Medicine Hannover. In cases of discrepancy between histopathologic diagnoses and those diagnoses obtained during routine histopathology and/or cytol-ogy, the specimens were critically re-evaluated by the same pathologist (M. H.-T.).
Assessment of cytological accuracy
Benign or inflammatory alterations were categorized as hyperplastic, extramedullary hematopoiesis, inflammatory, necrosis, hematoma, other degenerative or reactive-in-flammatory process and combinations of these. Cytologic and histopathologic diagno-ses were classified as “total agreement” when both agreed on tumor type (e. g., he-mangiosarcoma) or on the type of benign alteration (e. g., cytology: hyperplasia; histo-patholoy: nodular hyperplasia). Classsification as “partial agreement” required the cor-rect dignity (malignant vs. non-malignant) and, for example, was determined when ma-lignant diagnosis was made but cellular origin differed (e. g. hemangiosarcoma versus histiocytic sarcoma) or if cytologically malignant alteration was suspected and histo-pathologic examination revealed malignancy. Cases applied to category of partial agreement as well if benign disorders for example extramedullary hematopoiesis
(EMH) were seen in cytology und hyperplasia was found in histologic examination.
Samples of three FNP and six impression smears were considered as non-diagnostic and were excluded from evaluation. Interpretation of sample quality was done subjec-tively by using a scoring system between “0” and “+++” . Cellularity, preservation and staining quality of cells were assessed as well as blood contamination.
“0” represents specimens without diagnostically usable cell content, e. g. extremely low number of cells, destroyed cell material, and/or dominance of blood contamination,
“+” describes specimens with low number of representative intact and well stained cells, e. g. with high number of destroyed cells, “++” represents a moderate to high number of preserved and well stained cells, e .g. small areas with densely packed cells,
“+++” characterizes large areas with very high number (densely packed) of well pre-served and well stained cells with minimal blood contamination. In some cases, inter-mediate stages were used. Different cell populations were examined for their occur-rence and described between “+” to “+++” as well. For assessment of the influence of sampling on sample quality samples with different qualities were summarized to low (0/+, +), moderate (+/++ to ++) or high quality (++/+++ to +++) and for the influence of sample quality on the accuracy of diagnosis to the two categories low to moderate (0/+, +, +/++) or moderate to high quality (++, ++/+++, +++).
Statistical analysis
Data of nominal scaled variables (i. e. numbers with different degree of agreement of cytological and histological diagnosis between sampling techniques and sample qual-ites) were compared using Fisher`s exact test. Calculations were performed with the program SAS Enterprise Guide 7.1 (SAS Institute Inc. Cary, North Carolina, USA). A value of P ≤ 0.05 was considered significant.