Overall, DNA quantification highlighted comparable levels between cell pellet (HCC827 8.9 ng/µL, H358 115.0 ng/µL) and cell block (HCC827 7.5 ng/µL, H358 145.0 ng/µL) preparations from corresponding cell lines. In addition, the DNA integrity number (DIN) of cell pellet preparations was similar to those of HCC827-derived and H358-derived cell block samples (6.9 and 7.9 vs 6.9 and 8.9, respectively). RNA evaluation from HCC78 and H596 samples on TapeStation 4200 (Agilent) revealed that cell block sampling, but not cell pellet preparations, affected RNA yield (HCC78, 12.5 ng/µL vs 15.3 ng/µL; H596, 6.3 ng/µL vs 9.5 ng/µL, respectively) and index fragmentation (2.6 vs 4.0 and 1.6 vs 3.9). Hence, nucleic acid extraction and qualification of cell block specimens from mixed engineered cell lines were conducted. As expected, DNA and RNA yields were consistent with previous data: [(DNA 8.2 ng/µL, DIN 2.0; RNA 37.2 ng/µL, RIN 1.0) block 1; (DNA 6.1 ng/µL, DIN 2.9; RNA 25.4 ng/µL, RIN 1.0) block 2] (Supplementary Fig. 2).

NGS analysis was successfully carried out in all instances. Briefly, adequate technical parameters of DNA and RNA samples were obtained from semiautomated NGS systems [DNA sample: number of reads 566,693, mean read length 146, number of mapped reads 565,956, percent reads on target 99.6%, average reads per amplicon 13,415, uniformity of amplicon coverage 97.6%); RNA sample: number of reads 365,350, mean read length 117, mapped reads for SLC34A2(4)-OS1(32) 45,651, mapped reads for MET Δ exon 14 3066)]. Likewise, adequate parameters of DNA and RNA samples were obtained from fully automated NGS systems [DNA sample: number of reads 1,966,288, mean read length 97, number of mapped reads 1,951,334, percent reads on target 95.3%, average reads per amplicon 6828, uniformity of amplicon coverage 95.8%; RNA sample: number of reads 1,591,464, mean read length 101, mapped reads for SLC34A2(4)-OS1(32) 3327, mapped reads for MET Δ exon 14 349] (Table 1).

NGS analysis on two independent cell blocks from mixed engineered cell lines confirmed all the molecular alterations found in single-cell populations, showing no differences between semiautomated and fully automated NGS platforms (Supplementary Fig. 3A, 3B).

Overall, series of eight reference 5-µm slide sets, generated from previously validated cell block samples, were successfully shared with all participating institutions. Molecular results were shared with the coordinating institution within 30 working days, as recommended by the study protocol. Regarding the geographical distribution of the centers, 14 out of 24 (58.4%) were located in Northern Italy, 5 out of 24 (20.8%) in Central Italy, and 5 out of 24 (20.8%) in Southern Italy. Only one participating institution (ID#22) (4.2%) was unable to perform the molecular analysis and to convey the results to the coordinating center.

Regarding nucleic acids extraction, 21 out of 23 (91.3%) centers successfully carried out DNA purification, whereas 17 out of 23 (73.9%) participating centers successfully performed RNA extraction. Moreover, DNA and RNA purification was performed with manual strategies in 9 out of 21 (42.8%) and in 6 out of 17 (35.3%) institutions and with automated platforms in 12 out of 21 (57.2%) and in 11 out of 17 (64.7%) institutions. In a single case, neither DNA nor RNA purification was performed (ID#23). Owing to some technical issues, DNA purification failed in one automated test-based center. Furthermore, DNA and RNA quantification was performed by 20 out of 21 (95.2%) and 15 out of 17 (88.2%) participating institutions. In particular, a median DNA concentration of 3.3 ng/µL (range 0.1–10.0 ng/µL) was inspected with automated systems, whereas a median of 13.4 ng/µL (range 2.0–45.8 ng/µL) was inspected with manual procedures. Moreover, RNA concentrations of 5.7 ng/µL (range 0.2–11.9 ng/µL) and 9.3 ng/µL (range 0.5–18.0 ng/µL) were detected in the same settings (Table 2, Fig. 2).

DNA and RNA concentration by participating center. Scatterplots showing a DNA and b RNA concentration obtained by each participating center, colored by the extraction method (red, automatic procedure; blue, manual procedure)

As shown in Fig. 2, manual procedures yielded higher concentrations than automated procedures. Center ID#20 yielded higher DNA concentrations probably because it used all the available input slides for DNA isolation. Moreover, a comparative analysis between the DNA and RNA yields obtained either by automated or manual procedures suggested once more that manual extraction yields higher DNA concentrations (p = 0.038) than automated systems (Table 3).

However, no statistically significant difference was observed for RNA extraction (Table 4).

Overall, DNA-based biomarker analysis was successfully carried out by 22 out of 23 (95.6%) participating institutions, whereas RNA-based molecular analysis was achieved in 18 out of 19 (94.7%) centers equipped with all the necessary technical support; of note, RNA molecular analysis was successfully carried by ID#13 by using a fully automated platform without previous RNA extraction.

For the molecular analyses, participating centers adopted their own routine diagnostic method. For DNA-related biomarker testing, NGS, RT-PCR, and MassArray-based systems were employed in 14 out of 22 (63.6%), 5 out of 22 (22.8%), and 1 out of 22 (4.6%) institutions, respectively. Moreover, DNA-based biomarkers were tested with NGS plus RT-PCR platforms and pyrosequencing plus RT-PCR in two institutions (ID#8 and ID#20, respectively). Among the centers that adopted GS, one institution (ID#6) used two different NGS systems (hybridization and amplicon-based platforms) to validate the molecular results, whereas another institution (ID#5) used a commercially available NGS panel plus a customized NGS assay. Overall, hybridization-based and amplicon-based NGS systems were adopted in 12 out of 14 (85.7%) and 2 out of 14 (14.3%) participating centers, respectively. Among the institutions that implemented RT-PCR-based systems, 3 out of 5 (60.0%) adopted a semiautomated lyophilized system, whereas 2 out of 5 (40.0%) adopted a fully automated system. Concerning RNA-based biomarker testing, 7 out of 18 (38.9%) and 9 out of 18 (50.0%) participating institutions reported using NGS-based and RT-PCR-based solutions, respectively. Interestingly, two centers adopted a hybridization-based NGS system plus a semiautomated RT-PCR platform either to confirm their molecular results (ID#18) or to repeat their molecular analysis after failing with NGS (ID#17) (11.2%). As with DNA-based biomarkers, one participating laboratory employed two independent NGS systems (hybridization and amplicon-based platforms) to carry out RNA-based molecular analysis (ID#6). Overall, hybridization-based (5 out of 7, 71.4%) and amplicon-based strategies (2 out of 7, 28.6%) were assessed in the NGS group. Moreover, 2 out of 9 (22.2%) and 7 out of 9 (77.8%) fully automated and semiautomated automatized lyophilized RT-PCR systems were reported, respectively (Table 5).

Overall, KRAS exon 2 p.G12C and EGFR exon 19 p.E746_A750del hotspot mutations were successfully detected by all participating institutions. Interestingly, for the detection of KRAS exon 2 p.G12C, 14 out of 21 (66.6%), 4 out of 21 (19.0%), 1 out of 21 (4.8%), 1 out of 21 (4.8%), and 1 out of 21 (4.8%) institutions adopted NGS, RT-PCR, NGS plus RT-PCR, MassArray, and pyrosequencing systems, respectively; instead, for the detection of EGFR exon 19 p.E746_A750del, 14 out of 22 (63.7%), 6 out of 22 (27.3%), 1 out of 22 (4.5%), and 1 out of 22 (4.5%) adopted NGS, RT-PCR, NGS plus RT-PCR, and MassArray systems, respectively (Fig. 3). In this regard, the sequencing platforms yielded median mutant allelic fraction (MAF) values of 29.5% (range 14.0–52.0%) and 80.3% (range 61.0–92.0%) for KRAS p.G12C and EGFR p.E746_A750del mutations, respectively. Similarly, RT-PCR-based technologies yielded a median Ct of 29.9 (range 27.4–35.1) and 26.0 (range 24.9–27.5) for the same genetic alterations. For RNA-related molecular results, 15 out of 16 (93.7%) participating institutions detected MET exon 14 skipping mutation, whereas all institutions detected ROS1 gene fusion using RT-PCR. In addition, 13 out of 18 (72.2%) centers successfully detected both ROS1 rearrangements and MET Δ exon 14 skipping alterations. Overall, 7 out of 16 (43.8%), 8 out of 16 (50.0%), 1 out of 16 (6.2%), 5 out of 15 (33.3%), 8 out of 15 (53.4%), and 2 out of 15 (13.3%) institutions adopted NGS, RT-PCR, and NGS plus RT-PCR systems for the detection of ROS1 aberrant transcript and MET Δ exon 14 skipping molecular alterations (Fig. 4). In addition, in ID#17, RT-PCR was used after NGS failed to detect MET Δ exon 14 skipping molecular alterations. In this regard, NGS platforms generated median readcounts of 48,647.6 (range 84.0–138,166.0) and 2981.2 (range 37.0–8340.0) for ROS-1 rearrangement and MET exon 14 skipping, respectively. Similarly, RT-PCR based systems generated a median Ct of 28.6 (range 25.5–31.1) and 31.8 (range 27.8–35.0) for ROS-1 rearrangement and MET Δ exon 14 skipping alterations, respectively. Moreover, 13 out of 18 (72.2%) institutions also evaluated MET Δ exon 14 skipping molecular alteration on DNA samples. Interestingly, 2 out 13 (15.4%) suggested DNA-based molecular analysis as an alternative to the RNA testing strategy (Fig. 5). Of note, 11 centers successfully detected MET Δ exon 14 skipping molecular alteration using both DNA- and RNA-based detection strategies. In a single case (ID#11), ROS1 rearrangement was successfully detected without previous RNA-based molecular analysis. A list of additional molecular alterations reported by the participating institutions is provided in Supplementary Table 2.

Pie charts showing the different techniques used for analysis of DNA a KRAS and b EGFR hotspot mutations. NGS next-generation sequencing, PyroSeq pyrosequencing, KRAS Kirsten rat sarcoma viral oncogene homologue, EGFR epidermal growth factor receptor

Pie charts showing the different techniques used for analysis of RNA a ROS1 gene fusions and b MET exon 14 skipping. NGS next-generation sequencing, MET MET proto-oncogene, receptor tyrosine kinase, ROS1 c-ros oncogene 1

Bar chart showing the percentage of cases in which MET exon 14 skipping was detected using DNA-based technologies, RNA-based technologies, or both. MET MET proto-oncogene, receptor tyrosine kinase