Genomic DNA was isolated from formalin-fixed, paraffin embedded (FFPE) samples using 50-µm-thick sections of the tumour according to the manufacture’s protocol using ReliaPrep gDNA Tissue Miniprep system (cat. No. A2050, Promega, United States). Only paraffin blocks where cancer occupied >50% of the section were selected for DNA extraction. The isolated DNA was subjected to qualitative and quantitative assessment via spectrophotometry-based absorbance studies using Nanodrop, Biodrop Resolution, Cambridge, United Kingdom. Polymerase Chain Reaction (PCR) of GTF gene (exon 15) was carried out in a 25 µL total volume, containing 80 ng genomic DNA, 10 µM respective primers and using GoTaq colorless master mix (cat. No. M7142, Promega, United States).

The PCR program was set as: HotStarsTaq activation at 95°C for 5 min, followed by 40 cycles, each having denaturation at 95°C for 30 s, annealing at 55°C for 1 min and extension at 72°C for 45 s, and the final extension was 72°C for 10 min. The quality of the PCR product was assessed using electrophoresis by resolving a part of the PCR amplified product (3 µl) on 1.5% agarose gel for all considered samples. Clear band of expected size was observed in all samples which corresponded to the size of the desired amplicons. The PCR product was then cleaned up manually using polyethylene glycol (PEG) purification methods. The PEG purified PCR product was used for sequencing PCR with a total reaction volume of 10 µl. The sequencing PCR was set as: initial denaturation at 95°C, 5 min, followed by 35 cycles, each having denaturation at 95°C for 10 s, annealing at 55°C for 5 s and extension at 60°C for 4 min. The sequencing PCR product was cleaned up using 125 µM EDTA, 100% ethanol, and 3 M sodium citrate. Post-clean up, DNA was reconstituted in HiDi and incubated at 95°C for 5 min for denaturation and immediately put on ice to cool the PCR product. After this snap chilling, the samples were proceeded for sequencing. The sequencing was done using both forward and reverse primers (GTF2I_F: 5′-ATC​CCG​TAC​CCT​CTT​TTC​C-3′, GTF2I_R: 5′-AGA​CAA​GAG​TTC​AAC​AGG-3′) for greater accuracy and the results were analysed using SeqMan II software (DNASTAR).

For immunohistochemical staining of representative thymoma cases, paraffin embedded tissue was cut to 1 µm sections. The staining was performed using an automated slide preparation system (Benchmark Ultra, Roche Tissue Diagnostics, Ventana, Tucson, AZ, United States), a Ventana OptiView DAB IHC Detection Kit (Ventana, Tucson, AZ, United States)and commercially available monoclonal antibodies for CD20 (Clone L26, DCS, Hamburg, Germany), TdT (Clone SEN28, Leica Biosystems, Buffalo Grove, IL, USA) and CK19 (Clone RCK108, Agilent, Santa Clara, CA. United States).

Primary endpoint of this study was the GTF2Imutational status. Subgroup analyses and analysis of MG status were exploratory. Associations between clinico-pathological and molecular factors and geographical origin were assessed with Fisher’s exact test for categorical factors and with Wilcoxon test for age. Age-adjusted effects of clinical parameters were tested with the likelihood-ratio test of logistic regression models for origin between a model with predictors clinical parameter and age and a model with predictor age only. Heterogeneity of association between origin and mutational status across subgroups was analysed with the likelihood-ratio test between a model with origin and subgroup factor as predictors and a model including interaction term of both predictors. p-values below 0.05 were considered significant. Statistical software R 4.0 was used to carry out analyses.

A total of 138 resection specimens were retrieved of which 37 were Indian cases and 101 were German. Among these, 114 cases were thymomas consisting of 23 (20.2%) type A thymomas, 52 (45.6%) type AB, 17 atypical A/AB (14.9%) and 22 (19.3%) type B1-B3 thymomas. German thymoma cases (n = 77) included 33 female (42,9%) and 44 male (57,1%) patients in the age range 31–85 years (median age 65 years). Masaoka-Koga stage I was reported in 16/77 cases (20,8%), stage II in 38 cases (49,4%), stage III in 9 cases (11,7%) and stage IV in 14 cases (18,2%). The German cohort included 15 (19,5%) type A thymomas, 31 (40,3%) type AB, 15 atypical A/AB (19,5%) and 16 (20,8%) type B1-B3 thymomas. The remaining 24 German cases comprised 18 thymic carcinomas, 2 cases of sclerosing mediastinitis, 2 metaplastic thymomas, 1 thymolipoma and 1 thymic hyperplasia with lymphoepithelial sialadenitis (LESA)-like features (10).

All the 37 Indian cases included in this study were thymomas [8 type A (21.6%), 21 type AB (56.8%), 2 atypical A/AB (5.4%)and 6 type B1-B3 thymomas (16.2%)]. Among the 37 Indian cases, there was an equal gender distribution (19 cases or 51.4% were females). The age range in the Indian group was 20–73 years with a median age of 50 years; hence, the Indian cohort was significantly younger than the German counterpart (p < 0.0001, Table 1; Figure 1). Notably, even after excluding type B thymomas, this striking age difference remained in separate analyses of histotypes A and AB (p = 0.00097 and p = 0.0068, respectively; Supplementary Table S1). In both cohorts there were no significant differences between the age distribution of patients with and without MG. Masaoka-Koga stage distribution was as follows: 24 (64.9%) stage I, 9 (24.3%) stage II and 4 cases (10.8%) were in stage III.

Within defined WHO thymoma histotypes, we could not observe morphologic variations that could be assigned as specific to the Indian or German cohort, as confirmed by two experts in the field (AM and DJ). Representative histological and immunohistological features of thymoma cases included in this study are illustrated in Figure 2.

Also, no significant difference regarding the distribution of thymoma histotypes was evident (p = 0.16; Table 1). In age-adjusted analysis, there was a difference (p = 0.048), because age and histotype are strongly associated. On the other hand, we could detect a highly significant difference with respect to disease stage at clinical presentation, showing a clear majority of Indian patients at Masaoka-Koga Stage I (64.9% of all Indian thymoma cases) in comparison to only 20.8% of German patients at this early stage (p < 0.0001; Table 1). In our histotype A- and AB-dominated cohorts, this difference was most prominent for type AB (p = 0.006) followed by type A thymomas (p = 0.03; Tables 2 and 3).

In the German cohort, follow up information on myasthenia gravis status was available in 25/77 thymomas (of which 19 were type AB, 2 type A and 4 atypical A/AB) and 12 patients had myasthenia gravis (48% of patients with available information on myasthenia status). All tumours arising in myasthenia-positive German patients were diagnosed as type AB thymoma in Masaoka-Koga stage I or II (TNM: pT1a stage I).

Myasthenic status was known in 17 Indian patients, of these 14 had MG (82.3%) of which 6 (42.9%) had type A, 5 (35.7%) type AB thymomas and 3 cases (21.4%) type B thymomas. The single myasthenic patient with an atypical type A thymoma had recurrence of disease, and another patient with a “conventional” type A thymoma died due to myasthenic crisis. High risk features seen in the atypical A thymoma were increased atypia of the epithelial cells along with frequent mitotic activity and areas of necrosis.

Hence, a clear tendency towards significantly more frequent MG could be seen in the Indian cohort (82 vs. 48%; p = 0.0504; Table 1). When accounting for age differences between cohorts, no significant effect remained (p = 0.13; Table 1). No significant difference in association between geographical origin and MG status could be observed between age groups (p = 0.69; Figure 3). Intriguingly, all German patients with MG had thymomas of AB histotype, compared with only 35.7% of Indian patients (Tables 4 and 5).

The analysis of clinico-pathological factors between MG and non-MG patients (pooled German and Indian groups) showed significant differences in Masaoka-Koga stage between the two groups. Patients with MG presented at earlier stage (100% MG versus 56.2% non-MG pateints at Masaoka-Koga stage I or II; p = 0.0013; Table 6).

Out of the 114 thymomas, 3 Indian cases were excluded from mutational studies since the extracted DNA was of insufficient quality. Overall, mutation rates were similar in both cohorts (64 vs. 65%, p = 1.00, age-adjusted p = 0.26; Table 1). As expected, in general there was a strong association between histotypes A and AB and GTF2I mutation (p < 0.0001; Table 7). Out of a total of 89 type A/AB thymomas (including atypical forms), 71 (79.8%) were positive for the GTF2I mutation, including 78.6% (22/28) Indian and 80% (49/61) German cases. Overall, 20.2% (18/89) of types A/AB were negative for the mutation. Among the type B thymomas, overall prevalence of the mutation was 9.1% (2/22) of which all Indian B histotypes (n = 6) were negative, whereas 2 of 16 (12.5%) German B thymomas were positive, hence without significant difference (p > 0.05). In 25 German cases with known myasthenia gravis status, the GTF2I mutation was evident in 11/12 patients with myasthenia gravis (91,7%) and 10/13 myasthenia-free patients (76,9%). Out of 14 Indian cases with MG, 10 (71.4%) showed the GTF2I mutation; 1/3 patients (33.3%) without myasthenia harboured the mutation. There was no significant association of the GTF2I mutational status with the prevalence of MG overall and in both cohorts. No mutation was found in non-thymomatous samples, including 18 thymic carcinomas. Association between origin and mutational status was not significantly different between clinical subgroups (Figure 4).