Thymidine kinase 1 activity in serum and body cavity fluids of horses with lymphoma

Abstract

Background

The accuracy of serum thymidine kinase 1 (TK1_S) as a biomarker of lymphoma in horses is variable. The value of measurement in body cavity fluid (TK1_F) has not been established.

Hypothesis/Objectives

(1) TK1_F and TK1_S will be higher in horses with lymphoma compared to diseased controls (DCs); (2) TK1_F will be higher than TK1_S; and (3) there will be a strong correlation between both measurements.

Animals

Twenty animals in the lymphoma and 31 animals with nonneoplastic, gastrointestinal/respiratory disorders in DC group.

Methods

This was a cross-sectional study using prospectively collected data and samples obtained during routine diagnostic evaluation at referral hospitals. Thymidine kinase 1 was measured using chemiluminescence immunoassay and presented as median (and interquartile range). Nonparametric tests compared TK1_F and TK1_S between and within groups and Kendall’s Tau assessed correlation between TK1_F and TK1_S.

Results

Median TK1_F was 7.48 (2.20-48.23) U/L in horses with lymphoma vs 2.50 (0.59-4.83) U/L in the DC group (P = .007); median TK1_S was 4.26 (0.46-9.37) U/L vs 0.40 (0.40-0.75) U/L (P < .001), respectively. Both biomarkers showed good discrimination (area under the curve > 0.70) between lymphoma and the DC group, with optimal cutoff values of 5.7 and 2.5 U/L, respectively. Thymidine kinase 1 activity in body cavity fluid (3.62 [0.85-9.86] U/L) was significantly higher than TK1_S (0.50 [0.40-2.09] U/L; P < .001) and there was no significant correlation between them within groups.

Conclusions and clinical importance

Thymidine kinase 1 activity in body cavity fluid and TK1_S provide different information and should be interpreted independently. They proved useful in lymphoma diagnosis, but not as a standalone test.

Abstract Translated to German

Hintergrund

Die Zuverlässigkeit der Thymidinkinase-1-Aktivität im Serum (TK1_S) als Lymphom-Biomarker bei Pferden ist unterschiedlich. Die Messung in Körperhöhlenflüssigkeit (TK1_F) ist noch nicht etabliert.

Hypothesen/Ziele

(1) TK1_F und TK1_S sind bei Pferden mit Lymphom im Vergleich zu erkrankten Kontrolltieren (DC) erhöht; (2) TK1_F ist höher als TK1_S; (3) es besteht eine starke Korrelation zwischen beiden Biomarkern.

Tiere

Zwanzig Tiere in der Lymphomgruppe und einsinddreissig Tiere mit verschiedenen nicht-neoplastischen gastrointestinalen/respiratorischen Erkrankungen in der DC-Gruppe.

Methoden

Es handelte sich um eine Querschnittsstudie mit prospektiv erhobenen Daten und Proben, die im Rahmen der routinemäßigen diagnostischen Abklärung in Überweisungskliniken gewonnen wurden. TK1 wurde mittels Chemilumineszenz-Immunoassay gemessen und als Median (und Interquartilsabstand) dargestellt. Mittles nichtparametrischer Tests wurden TK1_F und TK1_S zwischen und innerhalb der Gruppen verglichen und Kendals Tau bewertete die Korrelation zwischen TK1_F und TK1_S.

Ergebnisse

Der Median von TK1_F betrug 7,48 (2,20-48,23) U/L bei Pferden mit Lymphom gegenüber 2,50 (0,59-4,83) U/L in der DC-Gruppe (P = 0,007) und der Medianwert für TK1_S betrug 4,26 (0,46-9,37) U/L gegenüber 0,40 (0,40-0,75) U/L (P < 0,001). Beide Biomarker zeigten eine gute Unterscheidungsfähigkeit (AUC > 0,70) zwischen der Lymphom- und der DC-Gruppe, mit optimalen Cut-off-Werten von 5,7 U/L und 2,5 U/L. TK1_F (3,62 (0,85-9,86) U/L) war signifikant höher als TK1_S (0,50 (0,40-2,09) U/L; P < 0,001), und es gab keine signifikante Korrelation zwischen denen innerhalb der Gruppen.

Schlussfolgerungen und klinische Bedeutung

TK1_F und TK1_S liefern unterschiedliche Informationen und sollten separat interpretiert werden. Sie waren in der Lymphomdiagnostik hilfreich, aber nicht als alleiniger Test geeignet.

Introduction

Lymphoma is the most common malignant neoplasia of the internal organs in horses and typically affects the abdominal cavity, and less frequently, the thoracic cavity and other organs.^1,2 Diagnosis remains challenging in horses, as clinical signs and laboratory findings are rarely specific.^1,3 Cytological or histopathological examination of affected tissue or associated fluid is necessary to confirm a neoplastic process.⁴ Although several circulating biomarkers are commercially available to assist with oncological investigations, they are often inaccurate due to the inconsistent nature of tumors and other factors affecting their levels.³

Thymidine kinase 1 (TK1) is an intracellular enzyme that leaks into the extracellular fluid during rapid cell proliferation. The measurement of TK1 activity in serum (TK1_S) is used as a biomarker in the oncology of small animals and humans.^5,6 In these species, many factors are considered when interpreting TK1_S, including age, accompanying inflammation, herpes virus infections, and liver and kidney disorders.³ There is some evidence that inflammation can cause an increase in equine TK1_S,^7,8 but the influence of the other factors remains unclear. Only a handful of studies have been undertaken in horses and demonstrated markedly divergent sensitivity and specificity of TK1_S as a lymphoma biomarker.^4,7,9

Given that lymphoma cells have a high proliferation rate, and they are usually surrounded by body cavity fluids, we hypothesized that (1) both TK1 in serum (TK1_S) and peritoneal or pleural fluid (TK1_F) will be higher in horses and ponies with lymphoma than in those with various gastrointestinal and respiratory disorders; (2) TK1_F will be higher than TK1_S; and (3) there will be a strong correlation between TK1_S and TK1_F. We aimed to reevaluate the performance of TK1_S as a lymphoma biomarker and investigate the potential of TK1_F as a novel approach in equine oncology.

Materials and methods

Please refer to Appendix 1.

Study design

Cross-sectional study with prospectively collected data.

Animals

The study considered horses and ponies exhibiting signs of chronic or recurrent gastrointestinal or respiratory disease, for which blood and peritoneal or pleural fluids were collected during the diagnostic work-up. Two groups were created: (1) lymphoma and (2) diseased controls with various gastrointestinal or respiratory disorders (DC). Animals were allocated to the first group if they were diagnosed with lymphoma affecting the bloodstream, or abdominal or thoracic cavity based on cytological or histopathological examination. Animals in the DC group were diagnosed with nonneoplastic diseases, such as inflammatory bowel disease (IBD), equine gastric ulcer syndrome, and recurrent large colon displacement. White blood cell counts in blood (WBC) and peritoneal or pleural fluid (WBC_F) were recorded if available.

Laboratory diagnostic

Both TK1_S and TK1_F were measured using chemiluminescence immunoassay (CLIA; Liaison XL, DiaSorin, Italy), which is commercially available (Laboklin GmbH, Bad Kissingen, Germany). The WBC and WBC_F were included in statistical analysis if they were measured at the primary institution using the same hematology analyzer (Heska Corporation, Loveland, Colorado, USA).

Statistical analysis

The power calculation showed that a sample size of 4 animals per group was required to detect differences in TK1_S between the lymphoma and DC groups, and total of 49 animals would be sufficient to assess the correlation between TK1_S and TK1_F. All statistical analyses were performed using IBM SPSS Statistics software (version 31.0.0.0 (117), Chicago, USA) and the significance level was set at P < .05. Normally distributed data (age) were presented as mean and SD, t-test was used to compare groups and Cohen’s d was calculated to assess the effect size. Not normally distributed data (TK1_S, TK1_F, WBC, WBC_F, and TK1_F/TK1_S) were presented as a median and IQR, Mann–Whitney U tests were performed to detect differences between the lymphoma and DC group, and r with 95% CIs were calculated to assess the effect sizes.

The area under the curve (AUC) with 95% CI and cutoff values associated with the highest Youden’s index were calculated based on a receiver operating characteristic (ROC) curve for the diagnosis of lymphoma using TK1_S and TK1_F, respectively.¹⁰

A Wilcoxon test was used to compare TK1_S and TK1_F. Kendall’s tau (τ) with 95% CI was calculated to investigate the correlation between TK1_S and TK1_F, TK1_S and WBC, and TK1_F and WBC_F.¹¹

Results

Descriptive statistics

There were 51 individual animals included in the study, whose descriptive statistics are summarized in Table 1. Forty-nine were presented between March 2023 and October 2025 and samples from 2 animals were retrieved from the sample archive stored at −80°C at the primary institution (collected in October 2021 and January 2022). The lymphoma group included 20 animals. Five out of 9 samples collected at the primary institution were stored at −80°C (TK1_S: 5.69; IQR, 0.52-40.70 U/L; TK1_F: 6.29; IQR, 0.40-58.10 U/L). Seven serum samples (TK1_S: 1.07; IQR, 0.68-9.97 U/L) and 9 body cavity fluid samples (TK1_F: 7.26; IQR, 1.86-21.90 U/L) obtained from other clinics underwent freezing at −20°C for variable periods ranging from days to months. The samples were cooled and shipped directly to the laboratory in the remaining lymphoma cases (TK1_S: 4.43; IQR, 0.40-6.12 U/L; TK1_F: 31.73; IQR, 5.68-101.00 U/L).

Table 1.

Descriptive statistics of animals included in the lymphoma group and the DC group with various gastrointestinal and respiratory diseases, comparing their sex, breed, presenting clinical signs, origin, body cavity fluid type, and sample processing protocol. Both the number of observations (n) and the percentage (%) are provided.

	Lymphoma (n = 20)	DC (n = 31)
Sex, n (%)
Mare	9 (45)	16 (52)
Gelding	10 (55)	15 (48)
Stallion	1 (5)	0
Breed, n (%)
European Warmblood	13 (65)	15 (48)
American Quarter Horse	0	3 (10)
Thoroughbred	1 (5)	1 (3)
Pony	3 (15)	4 (13)
Draft horse	1 (5)	3 (10)
Other	2 (10) one each of Noriker, Haflinger	5 (16) one each of Standardbred, Curly Horse, Irish Tinker, Konik and Icelandic Horse
Clinical signs, n (%)
Weight loss	9 (45)	11 (35)
Recurrent or persistent fever	7 (35)	0
Anorexia	6 (30)	1 (3)
Lethargy	4 (20)	0
Chronic diarrhea	3 (15)	5 (16)
Respiratory distress	3 (15)	1 (3)
Recurrent colic	2 (10)	25 (81)
Other	4 (20) epistaxis, vestibular syndrome and facial paralysis, weakness, loss of performance	0
Origin, n (%)
Primary institution	9 (45)	31 (100)
Other institutions	11 (55)	0
Body cavity fluid, n (%)
Peritoneal	13 (65)	31 (100)
Pleural	7 (35)	0
Sample processing, n (%)
Cooled, measured directly   • Serum   • Body cavity fluids	8 (40) 6 (30)	2 (6) 2 (6)
Stored at −20°C   • Serum   • Body cavity fluids	7 (35) 9 (45)	0 0
Stored at −80°C   • Serum   • Body cavity fluids	5 (25) 5 (25)	29 (94) 29 (94)

Abbreviation: DC = diseased control.

Weight loss, recurrent or persistent fever, and anorexia were the most frequent presenting clinical signs (Table 1). Lymphadenopathy (8/20) was a common finding during physical examinations, followed by ventral edema (4/20) and distended jugular veins with venous pulsation extending above the lower third of the neck (2/20). Ultrasonographic examinations often revealed thoracic effusion (6/20) and, less frequently, ascites (2/20), splenomegaly (1/20), and abdominal mass (1/20). Ancillary laboratory diagnostic detected absolute lymphocytosis (5/20), monoclonal gammopathy (3/20), thrombocytopenia (2/20), and immune-mediated anemia (1/20).

The final diagnosis was confirmed by cytological examination in 13 cases, including peritoneal (8/20) and pleural fluid (3/20), blood (1/20) and fine needle aspiration of an enlarged lymph node (1/20). Histopathological confirmation was obtained in 7 animals, including 4 with a full postmortem examination, 1 after an abdominal mass removal during laparotomy, and 2 after a biopsy of enlarged lymph nodes. Based on anatomical location, 13 lymphomas were classified as multicentric, 6 as alimentary and 1 as mediastinal. There were 7 pleural (TK1_F: 6.70; IQR, 2.61-7.69 U/L) and 13 peritoneal fluid samples (TK1_F: 9.86; IQR, 2.07-32.10 U/L). Immunophenotyping was performed in 2 samples, in independent laboratories, and revealed T-cell lymphoma in both (TK1_S: 71.40 U/L; 6.12 U/L). T-cell-rich, large B-cell lymphoma was suspected in 2 cases: one based on the histopathological structure of a lymph node (TK1_S: 0.68 U/L) and the other after PCR for antigen receptor rearrangement (PARR) examination (TK1_S: 6.02 U/L). Two pleural (TK1_F: 101.00 U/L, 4.68 U/L) and 1 peritoneal (TK1_F: 53.60 U/L) fluid samples from horses with lymphoma yielded negative results for equine herpes virus 5 (EHV-5). Only 2 animals were treated with corticosteroids before admission (TK1_S: 5.69 U/L; 0.89 U/L).

The DC group consisted of 31 animals. The majority of the samples were stored at −80‍°C (TK1_S: 0.40; IQR, 0.40-0.80 U/L; and TK1_F: 2.20; IQR, 0.58-4.38 U/L) and only 2 were measured directly (TK1_S: 0.40 U/L; 0.40 U/L; TK1_F: 10.90 U/L; 10.30 U/L). Recurrent colic and weight loss were the most frequently reported clinical signs (Table 1). There were 11 animals diagnosed with IBD, 5 with large colon displacement, 4 with equine gastric ulcer syndrome, 4 with large colon tympany, 3 with large colon impaction, and one each with sand enteropathy, ileum hypertrophy, and stricture of the small intestines. Only one animal exhibited signs of respiratory disease. It was diagnosed with multiple eosinophilic granulomas, resulting in partial tracheal obstruction. As lymphoma was initially suspected, peritoneal fluid was taken for cytology; therefore, the horse met the inclusion criteria set in the study. Animals with severe asthma were not included, since body cavity fluids were not routinely obtained.

The diagnosis of IBD was based on a histopathologic examination of a transendoscopic duodenum biopsy in 9 animals, 1 horse underwent laparotomy with full-thickness biopsies of intestines, and 1 had a postmortem examination. The other diagnoses were made based on routine gastroenterological or respiratory work-up, 4 underwent laparotomy and 1 postmortem examination.

The age was not significantly different between the lymphoma (15.5, SD 6.3 years) and DC group (13.6, SD 5.3 years; P = .286; Cohen’s d = 0.321). The WBC and WBC_F were only available for a limited number of cases, but no significant differences were found between groups (Table 2).

Table 2.

Comparison of thymidine kinase 1 measured in serum (TK1_S) and body cavity fluid (TK1_F), ratio of TK1_F to TK1_S (TK1_F/TK1_S), WBC and WBC_F between 2 groups: lymphoma and DC groups. The number of measurements is provided (n). All data are presented as medians (and IQRs). The independent samples Mann–Whitney U test was used to compare groups, and r with 95% CI was calculated to assess effect sizes.

	n	Lymphoma	n	DC	P value	r (95% CI)
TK1S (U/L)	20	4.26 (0.46-9.37)	31	0.40 (0.40-0.75)	<.001a	0.50 (0.44-0.56)
TK1F (U/L)	20	7.48 (2.20-48.23)	31	2.50 (0.59-4.83)	.007a	0.38 (0.12-0.59
TK1F/TK1S	20	3.8 (0.7-17.2)	31	3.3 (1.0-9.4)	.969	0.01 (−0.27-0.29)
WBC (G/L)	10	9.20 (6.45-10.53)	29	7.75 (6.52-9.13)	.258	0.18 (−0.10-0.43)
WBCF (G/L)	8	1.81 (1.17-8.16)	27	1.17 (0.57-2.65)	.154	0.24 (−0.04-0.48)

Abbreviations: DC = diseased control; TK1_F = thymidine kinase 1 activity in body cavity fluids; TK1_S = thymidine kinase 1 activity in serum; WBC = white blood cell counts in blood; WBC_F = white blood cell counts in peritoneal or pleural fluid.

^aStatistically significant.

Thymidine kinase 1 measurement in serum—TK1_S

The TK1_S was significantly higher in the lymphoma group than in the DC group (P < .001; Figure 1). Based on the ROC curve, the AUC for differentiating animals with lymphoma from the DC group using TK1_S was 0.782 (95% CI, 0.641-0.923; P < .001), indicating good discrimination according to Simundic¹⁰ (Figure 2). The highest Youden’s index of 0.55 indicated the cutoff value of 2.5 U/L. Sensitivity, specificity, and accuracy for this and the currently recommended cutoff value (3.0 U/L) were similar (Table 3).

Figure 1.

Scatterplots comparing the activity of thymidine kinase 1 (U/L) in serum and in body cavity fluids between the lymphoma group and the DC group. Dotted horizontal line indicates the calculated cutoff values of 2.5 and 5.7 U/L, respectively. Abbreviation: DC = diseased control.

Figure 2.

ROC curves for TK1_S and TK1_F for discerning animals with from those without lymphoma. The AUC is 0.782 and 0.726, respectively, indicating good discriminatory ability. Abbreviations: AUC = area under the curve; ROC = receiver operating characteristic; TK1_F = thymidine kinase 1 activity in body cavity fluids; TK1_S = thymidine kinase 1 activity in serum.

Table 3.

Table summarizing the sensitivity, specificity and accuracy for different cutoff values for TK1_S and TK1_F. The cutoff values of 2.5 U/L for TK1_S and 5.7 U/L for TK1_F were based on the highest Youden’s indexes, and 3.0 U/L is the currently recommended cutoff value for TK1_S.

Positive result	Sensitivity (%)	Specificity (%)	Accuracy (%)
TK1S ≥ 2.5 U/L	55	100	82
TK1S ≥ 3.0 U/L	50	100	80
TK1F ≥ 5.7 U/L	65	81	75
TK1S ≥ 2.5 U/L or TK1F ≥ 5.7 U/L	80	81	80
TK1S ≥ 3.0 U/L or TK1F ≥ 5.7 U/L	75	81	78

Abbreviations: TK1_F = thymidine kinase 1 activity in body cavity fluids; TK1_S = thymidine kinase 1 activity in serum.

No relevant correlation between TK1_S and WBC was detected overall (τ = −0.054; 95% CI, −0.266 to 0.164; P = .659), in the DC (τ = −0.165; 95% CI, −0.402 to 0.092; P = .252) or in the lymphoma group (τ = −0.024; 95% CI, −0.503 to 0.446; P = .926).

Thymidine kinase 1 measurement in body cavity fluids—TK1_F

The TK1_F was significantly higher in the lymphoma compared to the DC group (P = .007; Figure 1; Table 2). Based on the ROC curve, the AUC for differentiating animals with lymphoma from the DC group using TK1_F was 0.726 (95% CI, 0.572-0.880; P = .004), indicating good discrimination¹⁰ (Figure 2). The highest Youden’s index of 0.456 indicated the optimal cutoff value of 5.7 U/L (Table 3).

There were weak correlations between TK1_F and WBC_F overall (τ = 0.371; 95% CI, 0.156-0.553; P = .002) and in the DC group (τ = 0.313; 95% CI, 0.053-0.532; P = .024), but not in the lymphoma group (τ = 0.265; 95% CI, −0.360 to 0.725; P = .373).

Associations between TK1_S and TK1_F

The TK1_F (3.62; IQR, 0.85-9.86 U/L) was significantly higher than TK1_S (0.50; IQR, 0.40-2.09 U/L; P < .001) and the difference was also statistically significant when the lymphoma (P = .014) and the DC group (P < .001) were tested separately (Table 2).

There was only a weak positive correlation between TK1_S and TK1_F (τ = 0.260; 95% CI, 0.077-0.426; P = .013). The correlation was not statistically significant when groups were tested separately (lymphoma: τ = 0.183, 95% CI, −0.138 to 0.469; P = .277; DC: τ = 0.233, 95% CI, −0.012 to 0.452; P = .101).

The combined use of TK1_S and TK1_F to diagnose lymphoma was investigated. A TK1_F/TK1_S ratio showed no discrimination capability between the lymphoma and DC group (AUC = 0.497; 95% CI, 0.319-0.674; P = .972). The test sensitivity improved to 80% with a specificity of 81%, when the positive results were defined as TK1_S ≥ 2.5 U/L or TK1_F ≥ 5.7 U/L and negative results as TK1_S < 2.5 U/L and TK1_F < 5.7 U/L (Table 3).

Discussion

This study demonstrates that TK1 activity measured by the chemiluminescence immunoassay in both serum and body cavity fluids offers a good discrimination between horses and ponies with lymphoma from those with various gastrointestinal and respiratory diseases. This is consistent with our initial hypothesis but contrary to the current literature.⁷ Our results further demonstrate that the activity of TK1 is significantly higher in body cavity fluids compared to serum; however, the 2 measurements do not correlate with each other within groups. This suggests that each measurement provides distinct information and should be interpreted independently. Defining a positive result as either TK1_S or TK1_F equal to or above the respective cutoff value enables more lymphoma cases to be captured. This approach is more sensitive than using the 2 measurements separately, while retaining a specificity of over 80%.

Thymidine kinase 1 measurement in serum—TK1_S

On the one hand, our findings are similar to the results of others who independently showed that horses with lymphoma have significantly higher TK1_S activity compared to healthy controls and horses with inflammatory disorders.^4,9,12 Our results contrast with others who did not detect such a difference.⁷ Caution should be taken when comparing the results of different TK1 assays as they use different substrates.⁵ There is a strong correlation between radioenzymatic and CLIA assays in dogs⁶ and humans,¹³ but a similar comparison has not been performed in horses. The radioenzymatic method used by Larsdotter et al.⁴ and the tritium-labeled thymidine technique described by Wang et al.⁹ and Unger et al.¹² had a good to an excellent discrimination capability between horses with and without lymphoma. The CLIA used in our study and by Moore et al.⁷ captures only 55% of the TK1 activity measured with tritium-labeled thymidine,⁹ which can partially explain the poorer performance. Despite this certain limitation, CLIA does not involve radioactivity, which makes it safer for clinical use⁵ and is currently the only commercially available test.

Despite using the same assay, Moore et al.⁷ reported the poor discrimination capability of TK1_S. The discrepant outcomes of our studies might be caused by different inclusion criteria. We did not include cases of cutaneous lymphoma in our study in contrast to others,^4,7,9,12 where it could have contributed to lower mean TK1_S values in lymphoma groups. Despite bearing a risk of systemic progression, cutaneous lymphoma is usually less aggressive^1,14 and causes lower TK1_S in dogs.¹⁵

Similar to our study, Moore et al.⁷ considered differential diagnoses for alimentary lymphoma in the control group. However, unlike us, they included animals with other internal tumors, severe liver disorders, and acute inflammatory conditions, many of which had high TK1_S. This probably led to a lack of statistical difference in TK1_S between lymphoma and non-lymphoma groups. Comparing the 2 studies encourages further investigation into possible TK1 release from other dividing cells, including different neoplastic cells, hepatocytes, and leukocytes, in the course of inflammation or IBD, as described in other species.^3,7,16–19

Elevated TK1_S occurs in horses with hemopoietic tumors other than lymphoma, such as myelomonocytic leukemia,⁸ but also with non-hemopoietic internal tumors, such as pheochromocytoma, osteosarcoma, leiomyoma, and thymoma.^4,7,12 We diagnosed a 19-year-old Standardbred gelding with thymoma during a postmortem examination at our institution which had TK1_S of 8.51 U/L (not included in the study). Furthermore, there is evidence that TK1_S might be elevated in horses with skin neoplasia, including sarcoid, squamous cell carcinoma, and melanoma.¹² According to current literature, TK1_S should not be considered a lymphoma-specific biomarker in horses, and should be more accurately described as a “proliferation marker.”

The influence of hepatic disease on TK1_S has not been investigated in horses and was beyond the scope of this study. The liver values available were incomplete and measured using different assays, so a meaningful statistical analysis was not possible. Nevertheless, liver disease was not the final diagnosis for any of the animals included in our study. Therefore, we do not expect hepatic-derived TK1 to have influenced our results. There is a strong correlation between TK1_S and AST, ALT and LDH in people with viral hepatitis¹⁶; therefore, TK1_S measurement in horses affected by equine hepacivirus would be of particular interest.

We compared the WBC between the lymphoma and DC group to further investigate the influence of inflammation on our results, but detected no group difference and no correlation between the WBC and TK1_S. This suggests that leakage of TK1 from leukocytes was unlikely to be contributing to TK1_S in our study. However, the findings need to be interpreted with caution due to the low sample size, subsequent lack of differentiation between neutrophil and lymphocyte counts, and lack of animals with severe inflammation included in our study. Furthermore, current research on dogs reveals a correlation between TK1_S and acute phase protein concentrations rather than inflammatory cell counts.^19,20 Unger et al.¹² detected no correlation between the TK1_S and serum amyloid A concentration in their equine cohort; however, the highest TK1_S levels were reported in horses with severe inflammation in the tumor-free control group.

In conclusion, the absence of animals with other internal tumors, liver disorders, and inflammation in our study might explain the excellent specificity. Although the sensitivity of TK1_S around 50%-55% (depending on the selected cutoff value) seems unsatisfying (Table 3), it is similar to that reported in people.²¹ There are several factors that might affect the proliferative activity of lymphoma and subsequently limit the increase in TK1_S. These include certain drugs, tumor type and stage,⁷ and co-infection with EHV-5.³

Corticosteroids might limit TK1_S increase by inhibiting cell proliferation and inflammation.^4,7 Lack of treatment details (drug, dosage, and treatment duration) and the small number of animals treated prevented further conclusions in our study.

In agreement with current literature,^1,2 multicentric lymphoma was also the most common type in our study. A comparison between animals with different tumor topographies was not possible due to the small number of cases involving only the mediastinum or alimentary tract. Whether there are differences in TK1_S between immunophenotypes also remains to be investigated as ancillary diagnostics, such as immunohistochemistry, are still not routinely performed in equine patients. The PARR, utilized in one case in our study, is not validated in horses,²² and reports of its use are scarce.²³ Nevertheless, both T-cell and T-cell-rich, large B-cell lymphoma detected in our study are considered the most common in horses.^1,2

Several herpesviruses stimulate lymphoproliferation in different species, for example, Marek disease virus in poultry and Epstein–Barr virus in people.²⁴ A few authors suggested an association between lymphoma and EHV-5 infection in horses.^24–27 In our study, all body cavity fluids tested for EHV-5 returned negative results, but the corresponding TK1_F measurements were above the cutoff value. Nevertheless, given that tissue biopsies, rather than fluids, were examined in previous reports,^25–27 infection with EHV-5 cannot be ruled out.

Thymidine kinase 1 measurement in body cavity fluids—TK1_F

Analyzing body cavity fluids alongside blood is a common practice in equine medicine, but the measurement of TK1 activity in peritoneal and pleural fluid is a novel approach in lymphoma diagnostics. Thymidine kinase 1 activity in body cavity fluid was significantly higher than TK1_S, indicating the need for the establishment of independent reference ranges. Although there was no obvious difference in TK1 between peritoneal and pleural fluids, we cannot rule out the possibility that both fluids require their own reference ranges.

The TK1_F offered a good discrimination between animals with and without lymphoma in our study. It was equal to or above the cutoff value (≥5.7 U/L) in 11 animals with normal TK1_S activity (<2.5 U/L), including 5 lymphomas, 4 IBD, 1 large colon tympany, and 1 displacement. This explains the slightly higher sensitivity of TK1_F compared to TK1_S but lower specificity (Table 3), indicating that something other than lymphoma might be a source of TK1 in body cavity fluids. Based on our data, false positive results might occur due to the proliferation of leukocytes in the course of IBD or an inflammatory process in the abdomen. Further investigation of their influence on both TK1_S and TK1_F is necessary as IBD is the most common differential diagnosis to alimentary lymphoma in horses.¹

We have found that higher TK1_F correlated with high WBC_F, suggesting that the biomarker might be partially affected by the turnover of leukocytes. If this proves correct on a larger group of horses, including those with a high WBC_F due to peritonitis (or pleuritis), the risk of lymphoma overdiagnosis would be minimized if the TK1_F was adjusted for the number of leukocytes in the peritoneal fluid (TK1_F-to-WBC_F ratio). A similar concept has been described in dogs, where the neoplasia index correcting TK1_S for C-reactive protein was introduced.²⁰

Associations between TK1_S and TK1_F

Although a weak overall correlation was detected between TK1_S and TK1_F, none was found within the lymphoma and DC groups. This indicates that the results probably reflect group differences rather than a true correlation between the 2 biomarkers.

A significant difference between TK1 in serum and body cavity fluid, and absence of their correlation suggest a compartmentalized production or different clearance mechanisms. Significantly higher TK1_F than TK1_S values support our hypothesis that TK1 first leaks into body cavity fluid during proliferative processes in the abdomen or thorax. This highlights that both biomarkers provide distinct information and should be interpreted independently.

It seems reasonable to favor the TK1 measurement in the fluid that is primarily affected by the disease, for example, serum in animals with the presence of neoplastic cells in the bloodstream, pleural fluid in animals with thoracic masses, peritoneal fluid in animals with suspected alimentary lymphoma and, potentially, even cerebrospinal fluid in the neurological form. All animals with lymphocytosis in our study had elevated TK1_S (7.55; IQR, 6.10-9.97 U/L), but only 4/7 animals with thoracic involvement had elevated TK1_F in the pleural fluid (6.70; IQR, 2.61-7.69 U/L), and 8/12 lymphoma cases with abdominal involvement had TK1_F in the peritoneal fluid above the cutoff value (6.29; IQR, 1.64-9.86 U/L). This might be caused by biomarker limitations, but might also reflect the difficulty in determining the primary tumor location and subsequent fluid selection. Therefore, it seems easier to recommend measuring TK1 in both blood and available body cavity fluid. Measuring TK1 across multiple samples might be valuable, particularly in multicentric lymphoma.

The combination of tumor markers into neoplastic panels is a common recommendation in human medicine.²¹ Given the low sensitivity of TK1_S and TK1_F when used independently, we evaluated the combined use of these 2 measurements. Our results did not support the implementation of a TK1_F/TK1_S ratio, which is commonly used with other measurements (eg, lactate or glucose concentration). Nevertheless, the sensitivity of 75%-80% (depending on the cutoff value selected for TK1_S) could be achieved based on our dataset when a positive result was defined as TK1_S or TK1_F equal to or above the respective cutoff values (Table 3). Although the sensitivity was higher than that of either biomarker alone, the specificity of 81% was lower than that of TK1_S alone (Table 3). While this approach could be a valuable addition to oncological investigations, it should always be interpreted in the clinical context and not be considered a standalone, screening, or tumor-specific test. Finally, these exploratory findings require validation in a larger cohort.

Limitations

Equine lymphoma is a rare disease and only 20 animals met the inclusion criteria during the study period. The initial power analysis for the TK1_S comparison between groups was based on conservative assumptions about variability, derived from a limited body of literature. We recognized this sample size was probably insufficient to account for the biological variability in horses and a recent study comparing similar groups indicated that 28 animals were required.¹² Given the actual sample size in our study, the power remained > 80% to detect large effect sizes in TK1_S (3 U/L difference, assuming the true SD is up to 2 U/L); however, power might be insufficient for smaller effect sizes. Although our results indicated a medium-to-large effect (Table 2), the uncertain influence of confounding factors such as lymphoma characteristics, liver function, inflammation, and sample handling, as well as the narrow spectrum of diseases represented in the control group, might have contributed to our study being underpowered.

Whether our results can be extrapolated to different lymphoma anatomical types (the multicentric form was overrepresented), immunophenotypes (only 3 results were available), and different animals’ ages (the cohort’s age was higher than the mean 10-13 years reported in the literature^1,28) remain to be further investigated. Another limitation is insufficient data on liver values and WBCs, which prevented us from determining whether hepatocytes and leukocytes impacted the TK1 levels in our study.

Although the lack of healthy controls in the DC group must be considered a limitation, this group was representative of the main differential diagnoses for lymphoma. This approach reflected a real-world scenario, where samples are obtained from diseased animals. Nevertheless, our cutoff values remain speculative until they are validated in a larger cohort that includes healthy animals, as well as animals with other tumors. Furthermore, pleural fluid was analyzed along with peritoneal fluid in our study, and we recognize the need to establish independent TK1 reference values in both fluids.

The storage and shipment of samples were not standardized due to the multicenter nature of this study and the inclusion of results obtained during routine investigations. Lymphoma samples were more frequently analyzed directly (ie, without a freezing period) or stored at −20°C than samples in the DC group. We cannot rule out the possibility that this was partly responsible for the difference in TK1_S and TK1_F between the groups. However, evidence from studies on companion animals suggests that TK1 activity remains stable for years if the serum is separated within 60 min and stored at −20°C or below.^20,29

Similar to other studies,^4,7 the majority of the animals in the lymphoma group were diagnosed based on a cytological examination. However, both false positive and false negative results have previously been reported when the cytology of the peritoneal fluid is utilized in the tumor diagnostics.^3,30 Furthermore, most IBD cases were diagnosed based on the histopathological examination of transendoscopic duodenal biopsies. These samples are usually superficial and provide little architectural detail.³¹ There are also no standardized interpretation guidelines for equine specimens, which makes them unreliable for discerning alimentary lymphoma from lymphoplasmacytic IBD.³¹ Full-thickness biopsies and immunohistochemistry are currently recommended,^1,32 but were not performed in our study. Some authors suggest that IBD might precede the development of intestinal T-cell lymphoma in horses as it is described in small animals,^1,32 which makes the clear separation between these 2 diseases even more complex. Therefore, while we recognize that some animals might have been misdiagnosed and miscategorized in our study, we would like to emphasize that TK1 activity is unlikely to be an effective method of distinguishing between these 2 proliferative diseases.

Conclusion

We provided evidence that measuring the TK1 activity in serum using a chemiluminescence immunoassay could be a useful component of an oncological investigation in horses, particularly for diagnosing lymphoma. Our study also demonstrated a novel approach by measuring the TK1 activity in peritoneal and pleural fluids, which are routinely collected during a gastroenterological work-up of equine patients. Reference ranges specific to each body cavity fluid should be established for the TK1 activity in healthy horses, as the results obtained were higher than the TK1 activity in serum. As there was no correlation between TK1_S and TK1_F, they should be considered independent biomarkers. Measuring TK1 activity in both serum and body cavity fluid simultaneously offered the highest sensitivity with reasonable specificity to diagnose lymphoma in our dataset. Nevertheless, respective cutoff values remain speculative until tested on a population including healthy animals. Current evidence in equine medicine indicates that TK1_S and TK1_F should be considered proliferation rather than lymphoma markers and should not be used as standalone diagnostic tools.

Abbreviations

AUC

area under the receiver operating characteristic curve

CLIA

chemiluminescence immunoassay used to measure thymidine kinase 1 activity

DC

diseased control group, including horses and ponies with variable, nonneoplastic gastrointestinal and respiratory disorders

IBD

inflammatory bowel disease

ROC

receiver operating characteristic curve

TK1

thymidine kinase 1 activity

TK1_F

thymidine kinase 1 activity in body cavity fluid, such as peritoneal or pleural fluid

TK1_S

thymidine kinase 1 activity in serum

WBC

white blood cell counts in blood

WBC_F

white blood cell counts in peritoneal or pleural fluid

τ

Kendall’s τ

Appendix—Materials and methods

Study design

An observational, cross-sectional study with prospective data collection.

Animals

Only residual samples were collected for the purposes of this study, and the Berlin State Office for Health and Social Affairs did not consider the project to be an animal experiment (StN 038/23). All animal owners gave written consent to use their data for scientific purposes. The majority of animals included in this study were admitted to the primary institution (Freie Universität Berlin) between March 2023 and October 2025. Samples from lymphoma cases from equine clinics around Europe and material stored in sample archives were considered if the samples were submitted to and analyzed by the same laboratory.

The inclusion criteria were as follows: (1) horses and ponies older than 12 months; (2) chronic (>2 weeks duration) or recurrent signs of gastrointestinal or respiratory disease; and (3) collection of blood and peritoneal or pleural fluid indicated and performed during the diagnostic work-up. The clinical signs included weight loss, recurrent colic (more than 2 episodes requiring veterinary intervention), chronic diarrhea, recurrent or persistent fever (>38.5°C), chronic cough, or respiratory distress. The final diagnosis plus the age, breed, and sex of all animals were noted. White blood cell counts in blood (WBC), and peritoneal or pleural fluid (WBC_F) were recorded if available. Two groups were created based on the final diagnosis: (1) lymphoma and (2) diseased controls (DCs) with various, nonneoplastic gastrointestinal and respiratory disorders. Animals were allocated to the first group if they were diagnosed with lymphoma based on cytological or histopathological examination, and categorized as either alimentary or mediastinal according to the classification described by Durham et al.¹ The study also included multicentric lymphomas if they affected the abdominal or thoracic cavities, or the bloodstream (ie, absolute lymphocytosis due to lymphoma in the leukemic phase or lymphocytic leukemia). The immunophenotype (if available) and pretreatment with corticosteroids were noted. The DC group was intended to encompass common differential diagnoses for lymphomas affecting the alimentary system or the mediastinum, such as inflammatory bowel disease (IBD), equine gastric ulcer syndrome, recurrent large colon displacement and severe equine asthma.

Laboratory diagnostic

Both thymidine kinase 1 activity in serum (TK1_S) and thymidine kinase 1 activity in body cavity fluid (TK1_F) were recorded in all animals in which measurement was performed directly as part of a routine investigation. In these cases, serum and body cavity fluids were processed according to the respective institution’s internal protocols, shipped to the laboratory in a cooled container overnight, and analyzed on a regular business basis—typically within 1-2 days. In the remaining cases, residual serum (separated from whole blood by centrifugation within 60 min) and peritoneal or pleural fluid samples were collected in cryotubes without anticoagulants. These samples were frozen at −80°C within 24 h at the primary institution and were subsequently shipped on dry ice (within 24 h) every 3-5 months and analyzed immediately upon receipt. Samples collected at other institutions were stored at −20°C and shipped in a cooled container with express delivery (within 24 h).

Thymidine kinase 1 activity in serum and TK1_F were measured using a commercially available chemiluminescence immunoassay (Liaison XL, DiaSorin, Italy; Laboklin GmbH, Bad Kissingen, Germany). The internal analytical validation of the test for use in equine serum yielded an intra-assay variability of 5.33%-12.53% and inter-assay variability of 13.72%-14.16%. Based on linearity testing, the upper detection limit was found to be 100 U/L, with 0.5 U/L as the lower limit. A cutoff value of 3 U/L was recommended for differentiating between animals with and without lymphoma.

The WBC and WBC_F were included in the statistical analysis if they were measured at the primary institution using the same hematology analyzer (Heska Corporation, Loveland, Colorado, USA). Where immunohistochemistry was performed, antibodies against CD3 were used as a T-cell marker, and against CD20 (and PAX5 in one case) as a B-cell marker. A real-time PCR targeting glycoprotein B of equine herpesvirus type 5 (EHV-5) was performed on 3 body cavity fluid samples (Laboklin GmbH, Bad Kissingen, Germany).

Statistical analysis

A sample size of 4 animals per group was required to compare TK1_S between the lymphoma and DC groups, using an independent t-test with 80% power and a 2-sided α of 0.05. It was assumed that tumor-free animals with various gastrointestinal or respiratory disorders would have an average TK1_S of 1 U/L (SD: 1 U/L), and that horses with lymphoma would have an average of 4 U/L (SD: 1 U/L). A mean difference of 3 times the SD was assumed between groups, corresponding to a large effect size. These assumptions were made based on critically evaluated reports utilizing chemiluminescence immunoassays in horses,^7,8 and dogs,⁶ as well as personal observations. Further sample size calculation indicated that 49 animals were required to detect a strong correlation between TK1_S and TK1_F, assuming the Kendall’s tau (τ) correlation of 0.7 with a 95% CI 0.7 ± 0.2. We aimed to include 25 animals per group for both analyses.

All TK1 values outside the detection range were converted to absolute values to facilitate statistical analysis: TK1_S below 0.5 U/L was changed to 0.4 U/L in 25 cases, TK1_F below 0.5 U/L was changed to 0.4 U/L in 9 cases, and TK1_F > 100 U/L was changed to 101 U/L in 4 cases.

All statistical analyses were performed using IBM SPSS Statistics software (version 31.0.0.0 (117), Chicago, USA), and the significance level was set at P < .05. The normal distribution of the data was visually inspected. Data that were normally distributed (age) were presented as the mean and SD, and a t-test was used to compare groups. Cohen’s d was calculated to assess the effect size. Data that were not normally distributed (TK1_S, TK1_F, WBC, WBC_F, and TK1_F/TK1_S) were presented as median and IQR, Mann–Whitney U tests were performed to detect differences between the lymphoma and DC groups, and r (r = Z/√N; Z—standardized test statistic; N—total sample size) with 95% CI was calculated to assess the effect sizes.

Receiver operating characteristic (ROC) curves were prepared for the diagnosis of lymphoma using TK1_S, and TK1_F, respectively, and the area under the curve (AUC) with 95% CI was calculated and interpreted according to Simundic.¹⁰ Accuracy, sensitivity, and specificity were calculated for the cutoff values, which were selected based on the highest Youden’s index for TK1_S and TK1_F, respectively. The performance of the test when using TK1_S and TK1_F simultaneously (with a positive result defined as at least one variable being equal to or above the calculated cutoff value, and a negative result being defined as both variables being below the cutoff values) was evaluated against single variable.

A Wilcoxon test was used to compare TK1_S and TK1_F. Kendall’s τ was calculated to investigate the correlation between TK1_S and TK1_F, TK1_S and WBC, and TK1_F and WBC_F. The results were presented as Kendall’s τ with 95% CI and interpreted according to Schober et al.¹¹

Author contributions

Karolina Drozdzewska (Conceptualization, Formal analysis, Investigation, Methodology, Project administration, Writing—original draft), Sophie Burde (Investigation, Methodology, Writing—review & editing), Elizabeth Mueller (Methodology, Supervision, Writing—review & editing), Roswitha Merle (Formal analysis, Methodology, Writing—review & editing), and Heidrun Gehlen (Methodology, Supervision, Writing—review & editing)

Conflicts of interest

Sophie Burde and Elisabeth Müller are associated with Laboklin Laboratory for Clinical Diagnostics GmbH & Co. KG, which offers measurement of the thymidine kinase 1 activity described in the study. All authors declare that the interpretation of results and the preparation of the manuscript were conducted independently of the laboratory’s commercial interests. All other authors declare no conflicts of interest.

Funding

A discount on laboratory services was provided by Laboklin Laboratory for Clinical Diagnostics GmbH & Co. KG, Germany. Supported by Open Access Funds of Freie Universität Berlin.

Off-label antimicrobial declaration

The authors declare no off-label use of antimicrobials.

Institutional animal care and use committee or other approval declarations

The project was reviewed by the Berlin State Office for Health and Social Affairs and classified as not constituting an animal experiment (StN 038/23), as only residual samples were utilized. Written consent to use the data for scientific purposes was obtained from all animal owners, either at the clinic or when samples were submitted to the laboratory.

Human ethics approval declaration

The authors declare human ethics approval was not needed.