This prospective single-blinded, cross-over study (ClinicalTrials.gov ID: NCT05188131) was conducted at the Division of Endocrinology, Diabetology and Metabolism of the University Hospital “City of Health and Science of Turin” (Turin, Italy) between October and December 2021. The study was approved by the Local Ethics Committee (cod. 0096442, September 21st, 2021) and was in accordance with the principles of the Declaration of Helsinki. Written informed consent was obtained from all study participants.

We recruited 12 healthy subjects, 6 men and 6 women, eligible if they were aged 20–50 years old with a BMI between 18.5 and 25 kg/m2.

Exclusion criteria were as follows: history of gastric disease or peptic ulcer, previous gastro-intestinal bleeding or hemorrhagic diathesis, any ongoing acute pathologic processes or current pharmacologic treatment (including oral contraceptives), NSAIDs or acetyl-salicylic documented allergy, pregnancy or breastfeeding.

All participants underwent two test sessions, with an interval of at least 3 weeks: the drug (diclofenac 75 mg, diluted in normal saline solution 0.9% 100 cc) was administered intravenously (iv) in 15’ on 1 day and placebo (normal saline 100 cc) was infused iv in 15’ on another day. In female subjects, all evaluations were carried out in the follicular phase (within 10 days from the start of menstrual flow). In the week preceding the day of the procedure, subjects were asked to drink 1.5–2 L of water per day, with a controlled diet with 3–5 g of salt per day. On the morning of the test, all participants were fasting and not drinking for at least 8 and 2 h, respectively. Moreover, for the entire duration of the testing session all participants were not allowed to drink or eat.

The test session was divided into three observation times and, as a cross-over design, the same procedure and measurements were repeated identically in all the subjects on both occasions.

Day 0: subjects had to collect a 24 h urine sample (for fluid balance) and a salivary sample for the determination of cortisol and cortisone between 11 pm and 12am.

Day 1: on procedure day (between 7.30 and 8 am), subjects were evaluated at baseline with BIVA: height and weight were measured, and resistance (Rz), reactance (Xc) and phase angle (PhA) were registered; a fasting blood sample for MR-proADM, sodium (s-Na), copeptin, MR-proANP, ACTH and cortisol was drawn. A urine sample for osmolality (u-Osm), sodium (u-Na) and potassium (u-K) was collected as well. The drug or placebo was then administered and serial blood samples for s-Na and copeptin were drawn at times + 15’, + 30’, + 45’, + 60’, + 90’, + 120’ and + 240’. At each time point a second blood sample was taken and immediately centrifuged in order to extract serum and plasma. Thus, an aliquot of such materials was stored at − 80 °C for the subsequent determination of a simplified biochemical profile, including ACTH, cortisol, MR-proANP and MR-proADM. At the end of the test, subjects were re-evaluated with BIVA and another urine sample for u-Na, u-K and u-Osm was collected. Thereafter a new 24 h urine collection was started and at night subjects were asked to take a second salivary sample for the determination of cortisol and cortisone.

Day 3: 48 h after the test, all subjects were revaluated with BIVA and another urine sample for u-Na, u-K and u-Osm and a new blood sample for s-Na and copeptin were collected.

Copeptin, MR-proADM and MR-proANP: blood from an EDTA-containing tube was centrifuged at 4000 rpm for 5’ and a plasma aliquot was immediately frozen and stored at − 80 °C until analysis. Copeptin (pmol/L), MR-proADM (nmol/L) and MR-proANP (pmol/L) concentrations were then determined with the B.R.A.H.A.M.S. KRYPTOR compact PLUS (ThermoFisher Scientific, Hennigsdorf, Germany) automated method using the TRACE (Time-Resolved Amplifed Cryptate Emission) technique. The limit of detection (LOD) of the assay is 0.9 pmol/L for copeptin, 0.05 pmol/L for MR-proANP and 0.05 nmol/L for MR-proADM; intra- and inter-assay coefficients of variation were, respectively, < 7% and < 12% for copeptin and < 4% and < 11% for both MR-proANP and MR-proADM.p-Osm and u-Osm: the measurements of osmolality were performed by an automatic osmometer (Osmo Station OM-6050, ARKRAY Global, Kyoto, Japan) adopting the freezing point depression method as the measurement principle. LOD is 0 mOsm/kg with an intra-assay coefficient of variation < 1%.

ACTH and cortisol: ACTH (ng/L) levels were determined on plasma from EDTA test tubes, using chemiluminescent immunological methods (CLIA) on the LIAISON Analyzer platform (DiaSorin, Saluggia TO, Italy), whose sensibility was 1.6 ng/L; the intra-assay and inter-assay coefficient of variation ranged up to 4.9 and 8.8%, respectively. Serum cortisol levels (μg/L) were determined by a competitive electro-chemiluminescence immunoassay automated on Cobas e601 instrument (Roche Diagnostics GmbH, Germany). Analytical sensibility was 0.18 μg/L. Intra- and inter-assay precision ranged from 3.0 to 5.7% and from 2.4 to 6.2%, respectively.

Salivary cortisol and cortisone: salivary samples were collected using Salivette® (SARSTEDT, Nümbrecht, Germany). Cortisol and cortisone were determined using a LC–MS/MS analysis with the MassChrom® kit (Chromsystems Instruments & Chemicals GmbH, Gräfelfng, Germany). Nexera X2 UHPLC system (Shimadzu, Kyoto, Japan) was used for quantification, coupled with a triple-quadrupole mass spectrometer 4500MD (AB Sciex, Framingham, MA, USA). The sensibility for salivary cortisol is 0.28 µg/L, the intra- and inter-assay variation coefficient is 5.5 and 8.8%, respectively; cortisone sensibility is 0.55 µg/L, with intra- and inter-assay variation coefficients of 4.9 and 8.8%, respectively. Patients were instructed to soak the sample for 2–3’ and then put it in the plastic container at + 4 °C. Samples were collected at least 30’ before eating or drinking to avoid any contamination and patients were asked to have their teeth brushed at least 30’ before sampling. Smoking or eating licorice was prohibited and to ensure a valid sample collection, written instructions were given to patients.

Other biochemical analysis: every other routine biochemical determination was carried out through validated and automatized procedures in central Baldi&Riberi laboratory at the University Hospital "City of Health and Science of Turin".

BIVA was evaluated by an impedance vector analyzer with measurement frequency of 50 kHz ± 1% (BIA101BIVA®, Akern, Loc. Montacchiello, Pisa, Italy). Both bioimpedance parameters (Rz and Xc) were normalized according to the patients’ height (H) and plotted on a Rz/H and Xc/H graph (Biavector, Bodygram Plus® version 1.31). BIVA is a non-invasive technique that allows a reliable and reproducible assessment of the distribution of body fluids in several clinical settings [9], Rz reflecting conductivity through ionic solutions, Xc the impedance due to the membrane capacitance of metabolically active cells; finally, PhA represents a derived parameter, which expresses the ratio between intra- and extracellular fluid volumes. Furthermore, Biavector allows to compare the variations between repeated measurements on the same subject with the normal sex-specific ellipses of the general healthy population [10, 11]. Reliable thresholds for both overhydration and dehydration conditions have been previously identified at the lower and upper poles of the 75th sex-specific tolerance ellipse, respectively [11].

Shapiro–Wilk test was used to assess for normality. Normal variables were expressed as mean ± standard deviation. Non-normal variables that could be normalized after logarithmic transformation were then expressed as geometric mean and interquartile range. Unnormalisable variables were expressed as median and interquartile range.

Student’s t-test for paired samples or ANOVA test for repeated measurements were used to identify longitudinal differences in variables with a normal distribution. Wilcoxon and Friedman tests were used to identify differences in median values for variables that had a non-normal distribution.

To compare the profile of the analytes during infusion of placebo or diclofenac, another method was used, with the aim of comparing the trend during the whole test, rather than each single time of observation. A curve was plotted for the parameters of each subject in the two different experimental conditions and then the area under the curve (AUC) was calculated. A t-test for paired samples was then conducted between the AUC during placebo and the AUC during diclofenac.

With respect to BIVA, mean Biavectors’ displacement between consecutive observations was evaluated with the paired one-sample Hotelling's T2 test [10].

As this was a proof-of-concept study, no preliminary analysis for sample size and power were conducted. A cut-off p value < 0.05 was considered as statistically significant. Statistical analysis was performed using MedCalcTM® (Statistical Software version 18.11.3, MedCalc Software Ltd, Ostend, Belgium). Graphs and figures were realized with GraphPad PrismTM® (version 8.01; GraphPad Software Inc., La Jolla, California).