The study was designed as a prospective, observational trial. The study protocol was approved by the local ethics committee (BE 127/19). Written informed owner consent was obtained before enrolment of dogs in the study.

Healthy dogs and dogs meeting diagnostic criteria for nSIRS or sepsis presented to a small animal referral and teaching hospital between June 2020 and August 2021 were included.

All dogs were assessed at baseline (D1) via review of their medical history, physical examination including assessment of their mentation at presentation, hematology and blood biochemistry. In dogs with nSIRS and sepsis, blood lactate was also measured. Patient data of D1 (mentation, platelet concentration, albumin, glucose, lactate) then were used to calculate the shortened Acute Patient Physiologic and Laboratory Evaluation (APPLE_fast) score retrospectively. Healthy controls were sampled at presentation (D1) and on two consecutive days (D2; D3) for practical reasons. Dogs with nSIRS or sepsis were sampled at admission (D1), D2, D3 and D4 for measurement of pPCT, plasma CRP and complete blood count (CBC). Other diagnostic tests and all treatment decisions were at the discretion of the attending clinician.

Healthy dogs — Dogs were included if they had an uneventful medical history, had not received any medication in the previous 3 months except for preventative antiparasitic treatment, had a normal physical examination (eg. body temperature between 38.1°C and 39.2°C, heart rate between 60/min and 120/min depending on the size of the dog, respiratory rate <20/min), unremarkable CBC and plasma biochemistry profiles, including plasma CRP within the reference intervals (<10.5 mg/L).

Dogs with non-infectious systemic inflammatory response syndrome — A diagnosis of nSIRS was based on clinical criteria for nSIRS in dogs published by Hauptman et al. (13). Briefly, dogs were included if they fulfilled ≥ 2 of the following criteria at admission: body temperature < 38.1°C or > 39.2°C; heart rate > 120/min; respiratory rate > 20/min; white blood cell (WBC) count <6.0 × 10⁹/l (6 000/μl) or > 16.0 × 10⁹/l (16 000/μl), and percentage of band neutrophils >3% of the total WBC count, assessed by manual differential blood count. Additionally, diagnostic workup as described below revealed no evidence of bacterial infection; thus, bacterial infection was ruled out.

Dogs with bacterial sepsis — Bacterial sepsis was diagnosed in dogs with nSIRS and bacterial infection, which was confirmed by direct visualization of bacteria either on cytological or histopathological specimens, bacterial culture, or intraoperative evidence of a septic focus.

Samples were submitted to the on-campus bacteriology laboratory for aerobic and anaerobic culture as well as antimicrobial susceptibility testing.

Venous blood samples were obtained by venipuncture of the cephalic or saphenous veins using a 21-gage needle or from an intravenous catheter. In the latter case, a purge sample of 0.3 mL of blood was discarded before taking the actual sample. Blood was collected into ethylenediaminetetraacetic acid (EDTA) (Sarstedt AG, Nümbrecht, Germany) and lithium heparin tubes (Monovette, Sarstedt AG Nümbrecht, Germany).

D1 samples for hematology and blood biochemistry analyses as well as D2, D3, D4 samples for hematology were immediately analyzed using an automated analyzer (ADVIA 2120i, Siemens Healthcare, Zürich, Switzerland; Cobas c501, Roche diagnostics, Basel, Switzerland). D1 blood gas analysis for assessment of the lactate concentration was also performed directly after blood collection using an automated analyzer (RAPIDPoint 500e Blutgasanalysesystem, Siemens Healthineers International AG, Zürich, Switzerland) directly after sampling.

Lithium heparin samples at D1, D2, D3 and D4 for measurement of pPCT and CRP were centrifuged within 1 h of sampling for 10 min at 20°C with a relative centrifugal force of 3,000 × g. Directly after that, plasma was separated and aliquoted into five micro tubes of 0.5 mL (Sarstedt AG, Nümbrecht, Germany), which were then frozen at −80°C until the day of analysis. Frozen storage time ranged between 1 and 9 months.

Plasma PCT was measured batch wise using a previously extensively validated canine PCT ELISA (Biovendor, Asheville, USA), for measurement of pPCT in citrated and heparinized plasma samples (9, 14). To ensure that the assay produces reliable results in our hands, a dilutional study and investigation of intra- and inter-assay variability was conducted, showing acceptable dilutional linearity, as already shown in a previous study (15). All samples and standards were measured in duplicates. The assay was performed as done in a previous study using a 2-fold dilution of the samples (15) and following the manufacturers’ instructions. Samples were reanalyzed if the coefficient of variation (CV) between duplicates was above 15%. The upper limit of the assay was defined as the absorbance of the highest standard after having observed that linearity requirements were met. If the absorbance of a sample exceeded the absorbance of the highest standard, the sample was measured again at a higher dilution.

Intra-assay variability was calculated based on eight measurements of three samples (V1, V2, V3) as duplicates on one ELISA plate. Inter-assay variability was calculated based on measurements of three samples (V1, V2, V3) on all 12 ELISA plates showing intra- and inter-assay variabilities of 4 and 14%, respectively, (data not shown).

Samples from different days of the same dog were measured on the same plate except when further dilution was required.

Plasma CRP was measured batch wise using an automated analyzer Cobas c501 (Roche Diagnostics, Basel, Switzerland) with a previously validated CRP assay (Gentian Canine CRP; Scil Animal Care company, Viernheim, Germany) (16). The lower limit of quantification was set at 6.4 mg/L by the manufacturer; the upper limit of the reference interval was 10.5 mg/L.

White blood cell concentrations were obtained using an automated analyzer (ADVIA 2120i, Siemens Healthcare, Zürich, Switzerland) with the following veterinary software: ADVIA multispecies software, version 6.3.2. A manual differential blood count based on 200 leucocytes was performed by experienced laboratory technicians to calculate absolute segmented and band neutrophil concentrations.

Blood lactate concentrations were measured using an automated analyzer (RAPIDPoint 500e Blutgasanalysesystem, Siemens Healthineers International AG, Zürich, Switzerland).

Statistical analysis was performed using a commercial statistical software package (NCSS 2020, LLC. Kaysville, Utah, USA).

Statistical power and sample size analysis were performed using the following software: Sergeant, ESG, 2018. Epitools Epidemiological Calculators; Ausvet¹ based on pPCT concentrations of healthy dogs in the current cohort and previously reported concentrations in septic dogs (9). This was decided in order to get the best possible results, based on the fact that pPCT of the healthy dogs in the present study were higher than in previous studies. The mean of 110 pg/mL and variance of 1,600 pg/mL of the cohort of healthy dogs from the present study was used to detect a difference of 40 pg/mL in mean PCT between control and septic dogs with a power of 80% and a significance level of 0.05. Power analysis indicated that at least 28 subjects per group were needed.

Patient characteristics including sex, neuter status, age, body weight and baseline blood variables were analyzed using descriptive statistics. Shapiro Wilk testing indicated non-normal distribution for several continuous variables, which therefore were reported as median and interquartile range (IQR).

Day-to-day variability and inter-individual variability of pPCT were assessed in the group of healthy dogs by calculation of CVs (%) via the following formula: Standard deviation/Mean*100%.

Differences in concentrations of pPCT, markers of inflammation including WBC concentration, band neutrophil concentration, and CRP were compared between disease groups and between survivors and non-survivors in nSIRS and sepsis groups using the Friedman’s Rank Test with Bonferoni’s post-hoc adjustment for multiple comparisons, which resulted in a corrected significance level of p < 0.008.

Associations between D1 pPCT concentrations and antimicrobial pretreatment or APPLE_fast score were assessed using Kruskal Wallis Test.

Because of the non-normal distribution of the data, the correlation of pPCT and CRP, and the associations of pPCT with clinical and laboratory variables were assessed using the Spearman rank correlation test. R values between +/− 0.5 and +/− 1 were defined as a strong correlation, values between +/− 0.3 and +/− 0.49 were defined as a moderate correlation and values below +/− 0.29 were defined as a weak correlation.

Statistical significance was set at p < 0.05.

A total of 48 dogs were enrolled in the study: 17 dogs were diagnosed with bacterial sepsis, 16 with nSIRS, and 15 dogs were healthy. Population characteristics, baseline laboratory parameters, duration of hospitalization and outcome of dogs included in the study are summarized in Table 1. A positive outcome was defined as alive at discharge. Twelve dogs were crossbreds and 36 were purebreds belonging to 25 different breeds, the most common breeds being Border Collie (n = 3), German Shepherd dog (n = 3) and Labrador Retriever (n = 3). Of the 15 healthy dogs, 5 belonged to a cohort of military dogs. Of these 5 dogs, 2 were Malinois Shepherds, 2 were German Shepherds and 1 was a Dutch Shepherd. There were no significant differences between groups regarding age, neuter status, and body weight.

All non-surviving dogs in the study died or were euthanized because of intractable disease with poor prognosis. No dog was euthanized because of financial reasons. Non-surviving dogs were diagnosed with the following: septic peritonitis (n = 3), pneumonia (n = 2), pyometra (n = 1), intestinal sarcoma (n = 1).

Median APPLE_fast score was 21 (IQR 17–22); in septic dogs it was 21 (IQR 19–22) and in dogs with nSIRS 18.5 (IQR 17–21), the difference was not significant. Dogs with nSIRS and sepsis had significantly lower concentrations of plasma albumin compared to healthy dogs on D1.

In septic dogs, the most common sources of sepsis were septic peritonitis (n = 9) and pyometra (n = 3) and one of each of the following: pyothorax, pyelonephritis, abscess formation at the level of the neck. In two dogs in the sepsis group, bacterial pneumonia was strongly suspected but could not be confirmed definitively. As those dogs were not sufficiently stable to undergo bronchoalveolar lavage, diagnosis was based on clinical and radiographic signs as well as response to antimicrobial therapy.

Samples analyzed from dogs in the sepsis group included abdominal fluid, pleural effusion, abscess material, and tissue biopsies, depending on the clinical presentation. Isolated bacteria included Staphylococcus pseudintermedius (n = 1), Streptococcus equi subsp. Zooepidemicus (n = 1), Streptococcus anginosus (n = 1), Klebsiella pneumoniae (n = 2), Escherichia coli (n = 2), Enterobacter cloacae (n = 1), and Enterococcus faecium (n = 1).

Half of the dogs in the sepsis group (10/17; 58.8%) were treated before hospital admission with one or more of the following antimicrobials: ampicillin-sulbactam (n = 2), amoxicillin (n = 3), cefovecin (n = 1), enrofloxacin (n = 3), metronidazole (n = 4), marbofloxacin (n = 2), unknown antimicrobial (n = 1).

Dogs with nSIRS were diagnosed with steroid-responsive meningitis-arteritis (n = 4), heat stroke (n = 2), gastric dilation without torsion (n = 2) and one of each of the following: suspected acute toxic liver insult, intestinal sarcoma, Shar-Pei Fever, acute neutrophilic splenitis, primary immune-mediated polyarthritis, acute severe gastroenteritis, acute pancreatitis, gastric dilation volvulus.

Prior to admission, one third of the dogs in the nSIRS group (5/16; 31.3%) had received one or more antimicrobials, including amoxicillin-clavulanic acid (n = 1), enrofloxacin (n = 1), and an unknown antimicrobial (n = 4). The remaining dogs had not received antimicrobial treatment before presentation.

At presentation, median pPCT concentrations did not significantly differ between dogs with sepsis, nSIRS and healthy dogs (Figure 1A). Median pPCT at baseline was significantly higher in non-survivors (p < 0.05) than survivors in the sepsis group (Figure 1B). This is based on n = 5 non-survivors. In the nSIRS group only 1/16 dogs died, therefore the statistical comparison between survivors and non-survivors was not performed.

Plasma CRP concentrations at presentation were above the upper reference limit of the laboratory (10.5 mg/L) in all dogs with nSIRS or sepsis. Median CRP concentrations were not significantly different between sepsis and nSIRS groups (Figure 2A). There was no significant difference in median baseline CRP concentrations between survivors and non-survivors in the sepsis group (Figure 2B).

Comparison of median pPCT concentrations between D1, D2, D3 and D4 showed no significant differences between sampling time points in neither healthy, sepsis nor nSIRS groups (Figure 3A). Intra- and interindividual variabilities (CV_I and CV_G, respectively) were markedly higher in the sepsis (CV_I 31%/ CV_G 75%) and nSIRS groups (CV_I 31%/ CV_G 85%) compared to the healthy group (CV_I 15%/ CV_G 36%), with individual animals showing markedly increased pPCT on one (n = 2) or several (n = 1) time points. Two dogs in the nSIRS group with high pPCT (>400 pg/mL) were diagnosed with heat stroke and steroid responsive meningitis arteritis. In the dog in the sepsis group with a PCT of 529 pg/mL bacterial pneumonia was suspected; the dog died shortly after sampling.

Comparison of pPCT kinetics between sepsis survivors and non survivors showed low and stable median pPCT, whereas in the one non-surviving dog with sepsis, who survived beyond day 1, pPCT initially increased and then decreased over time (Figure 3B).

In dogs with nSIRS and sepsis, median CRP concentration progressively decreased over time. In dogs with sepsis, median CRP concentrations decreased significantly between day 1 and day 3 as well as day 1 and day 4 (D1/D3: p < 0.008; D1/D4: p < 0.008) (Figure 4A). In survivors, CRP concentrations decreased significantly between day 1 and day 4 (p < 0.008). Interpretation of kinetics in non-survivors is limited, as only one dog survived beyond the first day of sampling (Figure 4B).

In septic dogs, median pPCT concentration was significantly and strongly negatively correlated with WBC concentration (r = −0.6; p < 0.05) (Figure 5A). No statistically significant correlation was found between median pPCT and CRP concentrations (Figure 5B), band neutrophil concentrations or albumin.

In dogs with nSIRS, no statistically significant correlation was found between median concentrations of pPCT and CRP, WBC, band neutrophil concentration or albumin.

There was no association between pre-treatment with antimicrobials and median pPCT concentrations in septic dogs. No statistically significant correlation was found between the APPLE_fast score and D1 pPCT and CRP concentrations in septic dogs.