Clinical Relevance of Sensitive and Quantitative STAT3 Mutation Analysis Using Next- Generation Sequencing in T-Cell Large Granular Lymphocytic Leukemia  Thomas Kristensen, Martin Larsen, Annika Rewes, Henrik Frederiksen, Mads Thomassen, Michael Boe Møller  The Journal of Molecular Diagnostics  Volume 16, Issue 4, Pages 382-392 (July 2014) DOI: 10.1016/j.jmoldx.2014.02.005 Copyright © 2014 American Society for Investigative Pathology and the Association for Molecular Pathology Terms and Conditions

Figure 1 A: The four PCR primer pairs produced specific amplification of STAT3 exons 19 to 24. Both the no template control (NTC) and the shown sample were analyzed in duplicate (#1 and #2). B: Twofold dilution series of the purified and pooled PCR amplicons shown in A shows that the concentrations of each of the PCR amplicons are highly similar. Although the lengths of the four PCR amplicons are slightly different, no further adjustment to equimolar amounts was performed before pooling and NGS sample preparation. SM represents a DNA size marker with the indicated fragments lengths in base pairs. The Journal of Molecular Diagnostics 2014 16, 382-392DOI: (10.1016/j.jmoldx.2014.02.005) Copyright © 2014 American Society for Investigative Pathology and the Association for Molecular Pathology Terms and Conditions

Figure 2 Results of NGS and Sanger sequencing analysis of three STAT3 Y640F mutation-positive patients; patient 5 with 33.9% (A and D), patient 4 with 12.5% (B and E), and patient 10 with 2.5% mutation-positive alleles (C and F). Numbers on the y axis in A–C represent the percentage of the indicated nucleotide. Sanger sequencing in both the forward (upper panel) and reverse (lower panel) directions confirmed the Y640F mutation in one patient (D), produced a borderline result that was due to higher than normal background in the lower panel in one patient (E) and produced a clear mutation-negative result in one patient (F). Mutation-positive results that were not unequivocally confirmed by Sanger sequencing in both directions were subjected to a second NGS analysis using a new DNA extraction and the mutation was thereby confirmed in all cases. The Journal of Molecular Diagnostics 2014 16, 382-392DOI: (10.1016/j.jmoldx.2014.02.005) Copyright © 2014 American Society for Investigative Pathology and the Association for Molecular Pathology Terms and Conditions

Figure 3 STAT3 mutation levels in four paired PB (white bars) and BM aspirate (black bars) samples from three patients. The Journal of Molecular Diagnostics 2014 16, 382-392DOI: (10.1016/j.jmoldx.2014.02.005) Copyright © 2014 American Society for Investigative Pathology and the Association for Molecular Pathology Terms and Conditions

Figure 4 STAT3 mutation-positive allele fraction over time in PB in four mutation-positive patients. Two patients did not receive cytoreductive treatment (A, squares and diamonds, solid lines), and two patients received cytoreductive treatment (B, dotted lines) during the shown time course. Black symbols represent mutation-positive samples, whereas the white symbols in one of the untreated patients represent mutation levels below the cutoff for positivity. Of the patients receiving treatment, one patient (triangles) continuously received peginterferon alfa-2a from day 29 (arrow 1) after the initial sample. The other patient (circles) received methotrexate from day 92 (arrow 2) to day 264 (arrow 3). The patient then received cyclophosphamide from day 602 (arrow 4) to day 677 (arrow 5), peginterferon alfa-2a from day 678 to day 806 (arrow 6), and pentostatin from day 827 (arrow 7) to day 979 (arrow 8). Transformation to an ALK-negative anaplastic large-cell lymphoma was diagnosed at day 991 and the patient therefore received CHOP (cyclophosphamide, doxorubicin, vincristine, and prednisolone) chemotherapy regimen from day 999 (arrow 9) to day 1120 (arrow 10). The Journal of Molecular Diagnostics 2014 16, 382-392DOI: (10.1016/j.jmoldx.2014.02.005) Copyright © 2014 American Society for Investigative Pathology and the Association for Molecular Pathology Terms and Conditions

Figure 5 Frequency distribution of the error rates of bases with a Q-value ≥20 in exons 19 to 24 (+20 bp) of the STAT3 gene in PB samples from 20 healthy donors. Note the logarithmic scale on the y axis. The three possible mismatches represent individual data points. None of the 54.360 error rates included in the figure exceeded 1%, and the result thereby supports a threshold for positivity of 1% as used by the variant caller software in the present study. The Journal of Molecular Diagnostics 2014 16, 382-392DOI: (10.1016/j.jmoldx.2014.02.005) Copyright © 2014 American Society for Investigative Pathology and the Association for Molecular Pathology Terms and Conditions

Figure 6 Nucleotide-specific error rates of bases with a Q-value ≥20 in exons 19 to 24 (+20 bp) of the STAT3 gene in PB samples from 20 healthy donors. Results are presented as means ± SD. The Journal of Molecular Diagnostics 2014 16, 382-392DOI: (10.1016/j.jmoldx.2014.02.005) Copyright © 2014 American Society for Investigative Pathology and the Association for Molecular Pathology Terms and Conditions

Figure 7 STAT3 mutation analysis of a twofold serial dilution of a D661Y mutation–positive sample diluted into wild-type DNA. Results are depicted as the mutation-positive allele frequency as a function of the dilution grade: STAT3 mutation-positive result (black squares) and STAT3 mutation-negative result (white squares) as determined by SVC. The values of the mutation-negative samples represent the mismatch corresponding to the D661Y mutation that is below the cutoff for positivity at 1% but above the typical error rate of the G-to-T mismatch and therefore most likely represent the mutation being present at a low level close to the detection limit as expected. A slope of −1.03 was detected using linear regression analysis of the mutation-positive samples. The Journal of Molecular Diagnostics 2014 16, 382-392DOI: (10.1016/j.jmoldx.2014.02.005) Copyright © 2014 American Society for Investigative Pathology and the Association for Molecular Pathology Terms and Conditions

Figure 8 The percentage neoplastic T cells determined by flow cytometry versus the percentage of STAT3 mutation–positive alleles in different fractions of the same samples in eight patients (PB n = 3, and BM n = 5). A significant correlation was detected between the two variables (P = 0.02, R2 = 0.65). The depicted line represents the result of linear regression. The Journal of Molecular Diagnostics 2014 16, 382-392DOI: (10.1016/j.jmoldx.2014.02.005) Copyright © 2014 American Society for Investigative Pathology and the Association for Molecular Pathology Terms and Conditions