The Swedish cardiopulmonary bio-image study (SCAPIS) is a general population-based prospective study [20] conducted at six Swedish university hospitals in Gothenburg, Linköping, Malmö/Lund, Stockholm, Umeå and Uppsala, with each site recruiting participants from corresponding municipality areas. Between 2013 and 2018, over 30 000 participants aged 50–64 years were invited to a comprehensive examination. Over a period of two days, the participants underwent extensive imaging and functional studies of the heart, lungs and metabolism and completed extensive questionnaires including lifestyle and medication. The SCAPIS Leukocyte substudy was undertaken to assess the immune profile by flow cytometry in fresh whole blood samples. Between November 2015 and May 2018, 1 077 participants in SCAPIS Leukocyte were consecutively recruited from the SCAPIS Linköping cohort. We strived for gender balance but otherwise no exclusion criteria were applied except the inability to understand written and spoken Swedish for informed consent.

To obtain absolute counts of CD3 + , CD4+ and CD8+ T cells, EDTA-whole blood was stained with anti-CD3 FITC, anti-CD8 PE, anti-CD45 PerCP and anti-CD4 APC antibodies using BD Multitest Trucount tubes (Becton, Dickinson and Company, Franklin Lakes, US) according to the manufacturer’s instructions. Proportions of naïve T cells including naïve Treg cells within the CD4+ compartment, as well as naïve T cells within the CD8+ T cell compartment were determined by multi-colour whole blood staining using the following monoclonal antibodies: anti-CD3 PerCP (Clone:SK7), anti-CD4 PE-Cy7 (Clone: SK3), anti-CD8 APC-H7 (Clone: SK1), anti-CD45RA FITC (Clone: L48) and anti-CCR7 BV421 (Clone: 150,503). Four µL of each antibody were used to stain 50 µL EDTA-whole blood for 15 min in room temperature in the dark. Erythrocytes were lysed using 1 × FACS Lysing Solution for 15 min in room temperature. After incubation, cell supernatant was separated by two rounds of centrifugation (300 × g, 20ºC for 5 min). 2 mL and 500 µL of phosphate-buffered saline / 0.4% (w/v) human serum albumin solution was mixed with the supernatant after the first and second round of centrifugation, respectively. The cells were then analyzed on a FACS Canto II flow cytometer by using FACSDiva software (BD Biosciences). All reagents were purchased from BD Biosciences, Sweden. Naïve CD4+ and CD8+ T cell subsets were defined as CD45RA+CCR7+ cells within the CD3+CD4+ and CD3+CD8+ populations, respectively. Naïve Treg cells were defined as CD25++CD45RA+ within the CD3+CD4+ population, as described by Hellberg et al. [21]. The gating strategy for CD3+, CD4+ and CD8+ T cells is shown in Supplementary Fig. 1, and the gating strategy for naïve CD4 + T cells, naive CD8 + T cells and naïve Treg cells in Supplementary Fig. 2.

IL-6 and IL-18 in EDTA plasma samples were analyzed at SciLifeLab Affinity Proteomics Uppsala, Sweden, using the U-Plex Meso Scale Discovery (MSD) platform (Rockville, Maryland, USA), an electrochemiluminescence assay, according to the manufacturer’s instructions. Inter-assay coefficients of variations (CV) were 21.3% for IL-6 and 10,6% for IL-18.

In a subset of 55 consecutively recruited individuals, peripheral blood mononuclear cells were isolated using Ficoll gradient centrifugation and stored until use in liquid nitrogen. DNA purification was performed using QIAamp DNA Mini Kit (Protocol: DNA purification from blood or body fluids—spin protocol, Qiagen), according to the manufacturer’s instructions. Levels of TRECs were analyzed with droplet digital PCR (BioRad). For the amplification of TRECs a custom assay from BioRad was used containing the following sequences: forward primer 5’-CAC ATC CCT TTC AAC CAT GCT-3’ (900 nM), reverse primer 5’-GCC AGC TGC AGG GTT TAG G-3’ (900 nM), probe 5’-ACA CCT CTG TTT TTG TAA AGG TGC CCA CT-3’ (dye 5′6-FAM, quencher 3’ Iowa Black FQ, 250 nM). A reaction mix consisting of supermix (ddPCR SMX for Probes, no dUTP, cat no 1863023, BioRad), TREC assay (described above), distilled water and Hind III reaction enzyme (5 U/reaction, recombinant 10.000 units, cat no R0104S, BioNordika) was added to DNA (1.1 µg DNA per reaction). After incubation for 20 min, droplets were generated using an automatic droplet generator (QX200 Droplet Generator, BioRad) followed by PCR amplification (C1000 Touch Thermal Cycler, BioRad). After incubation at + 4℃ overnight, the plate was inserted into a droplet reader (QX200 Droplet Reader, BioRad) and the results were presented as copies/µL in the Quantasoft Software (version 1.7.4, BioRad). A positive control (a 29-year-old healthy volunteer) and a no template control was included in each run.

In total, 1 077 participants were included. Of these, 29 (2.7%) were excluded; nine (0.8%) had missing images, six (0.6%) had anterior mediastinal lymphadenopathy/tumor, five (0.5%) had previous surgery preventing evaluation, two (0.2%) had artefacts and seven (0.6%) could not be evaluated due to miscellaneous anatomical and technical reasons. Thus, data from in total 1 048 participants were included in the final analysis.

According to the SCAPIS protocol [20], a CT of thorax was performed using a dedicated dual-source CT scanner equipped with a Stellar Detector (Somatom Definition Flash, Siemens Medical Solution, Forchheim, Germany). All scans were acquired with a low radiation dose protocol at maximum inspiration without intravenous contrast. Image acquisition parameters were as follows: tube voltage 120 kV, automatic current modulation (CARE Dose4D; Siemens Medical Solutions) quality ref 25 mAs, pitch 0.9 and rotation time 0,5 s. Detector configuration was 128 × 0,6 mm. Reconstructions used were performed with an iterative reconstruction soft tissue kernel (SAFIRE, I31), matrix size 512 × 512, slice thickness 3 mm, increment 2 mm.

All scans were reviewed using a fixed window setting (WL = 50, WW = 350) on a picture archiving and communication system station. Evaluations were performed by a thoracic radiologist with 19 years´ experience in interpreting chest CT scans. Thymus fat and soft tissue content was visually analyzed and graded on a scale 0–3, as previously described [13, 15]: grade 0, complete fatty replacement, no identifiable soft tissue in the thymic bed; grade 1, predominantly fatty thymus; grade 2, approximately one-half fatty and one-half soft tissue attenuation thymus; and grade 3, predominantly soft tissue attenuation thymus (Fig. 1).

Representative CT images of each thymic score

Each thymic score (indicated at top left in the image) is represented by two different cases. Cases 1 and 2: Score 0, complete fatty replacement, no identifiable soft tissue in the thymic bed, Cases 3 and 4: Score 1, predominantly fatty thymus, Cases 5 and 6: Score 2, approximately one-half fatty and one-half soft tissue attenuation thymus, and Cases 7 and 8: Score 3, predominantly soft tissue attenuation thymus.

Quantitative measurements were performed in individuals with scores 1, 2 and 3, including thymic attenuation and thymic size (anteroposterior and transverse diameters, length and thickness of right and left lobes). An axial image with maximum anterior–posterior diameter of thymic gland was selected, and all following measurements and evaluations were performed in the same image [13, 22]. A region of interest for attenuation measurement in Hounsfield units (HU) was defined, encompassing thymic soft tissue, excluding as much adjacent fat tissue as possible [13, 22, 23]. The morphology was determined using the following categories: 1) pyramidal with convex margins, 2) pyramidal with straight margins, 3) pyramidal with concave margins, 4) round or oval, and 5) irregular [13]. If thymic lobes had different configurations, the dominant lobe was registered [13].

To assess intra- and inter-observer agreement of thymic scores, two thoracic radiologists reviewed 112 randomly selected CT scans. One radiologist scored the 112 cases twice and the second investigator scored the same 112 cases once. For those evaluations, the radiologists were blinded to SCAPIS data and previous radiological assessments.

Weight was measured with participants in light clothing, using calibrated scales. BMI was calculated by dividing the weight (kg) by the square of the height (m). Abdominal obesity was defined as a waist circumference ≥ 102 cm in men and ≥ 88 cm in women. Office blood pressure was calculated as the average of three measurements, after 5-min supine rest, measured in the right brachial artery.

Sensor-based sedentary behavior and physical activity patterns were derived from tri-axial accelerometers (ActiGraph LCC, Pensacola, FL, USA), as previously described [24]. The participants wore the accelerometers for seven days. Wear time was calculated as 24 h minus non-wear time. Participants with a minimum of 600 min of valid daily wear time for at least 4 days were included. Total physical activity was expressed in daily mean tri-axial vector magnitude counts per minute (cpm). Sedentary time was defined as < 200 cpm, low intensity physical activity as 200–2689 cpm, moderate intensity physical activity as 2690–6166 cpm, and vigorous physical activity as ≥ 6167 cpm.

A shorter form of the food-frequency questionnaire Meal-Q (MiniMeal-Q) was used to assess dietary intake [25, 26]. The daily intake of energy, macronutrients, fiber and micronutrients was retrieved by linking intake of food items and dishes assessed with MiniMeal-Q to the Swedish food composition database provided by the Swedish National Food Agency and calculated as the average intake of units per day.

Plasma levels of high-sensitivity C-reactive protein (CRP) were measured using an immunoturbidimetric assay with a Roche Cobas C502 analyzer (Roche Diagnostics, Scandinavia AB), with a CV of 2.2%. Total cholesterol, high-density lipoprotein (HDL) cholesterol, low-density lipoprotein (LDL) cholesterol, triglycerides and glycated hemoglobin (HbA1c) were measured in a fasting condition at study entry through standard analyses at an accredited laboratory. White blood cell differential counts in whole blood were determined by Cell-Dyn Sapphire™ (Abbot Diagnostics).

IBM SPSS Statistics (version 28) and GraphPad Prism (version 9.1.2) were used for analyses. Cohen´s weighed kappa coefficients were used to determine intra- and inter-observer agreements of thymic scores. The Kruskal–Wallis H test, alone or in combination with Dunn’s multiple comparisons test, or the Mann–Whitney test were used to compare characteristics between groups with different thymic scores. Spearman´s correlation was used to analyze univariate correlations between quantitative variables. Logistic regression was used to analyze which factors were associated with discrepancies between different thymic scores. Linear regression was used to determine independent determinants of CT attenuation and thymic size.

The data that support the findings of this study are available from the corresponding author upon reasonable request.