Clonality assessment and detection of clonal diversity in classic Hodgkin lymphoma by next-generation sequencing of immunoglobulin gene rearrangements

Clonality analysis in classic Hodgkin lymphoma (cHL) is of added value for correctly diagnosing patients with atypical presentation or histology reminiscent of T cell lymphoma, and for establishing the clonal relationship in patients with recurrent disease. However, such analysis has been hampered by the sparsity of malignant Hodgkin and Reed-Sternberg (HRS) cells in a background of reactive immune cells. Recently, the EuroClonality-NGS Working Group developed a novel next-generation sequencing (NGS)-based assay and bioinformatics platform (ARResT/Interrogate) to detect immunoglobulin (IG) gene rearrangements for clonality testing in B-cell lymphoproliferations. Here, we demonstrate the improved performance of IG-NGS compared to conventional BIOMED-2/EuroClonality analysis to detect clonal gene rearrangements in 16 well-characterized primary cHL cases within the IG heavy chain (IGH) and kappa light chain (IGK) loci. This was most obvious in formalin-fixed paraffin-embedded (FFPE) tissue specimens, where three times more clonal cases were detected with IG-NGS (9 cases) compared to BIOMED-2 (3 cases). In total, almost four times more clonal rearrangements were detected in FFPE with IG-NGS (N = 23) as compared to BIOMED-2/EuroClonality (N = 6) as judged on identical IGH and IGK targets. The same clonal rearrangements were also identified in paired fresh frozen cHL samples. To validate the neoplastic origin of the detected clonotypes, IG-NGS clonality analysis was performed on isolated HRS cells, demonstrating identical clonotypes as detected in cHL whole-tissue specimens. Interestingly, IG-NGS and HRS single-cell analysis after DEPArray™ digital sorting revealed rearrangement patterns and copy number variation profiles indicating clonal diversity and intratumoral heterogeneity in cHL. Our data demonstrate improved performance of NGS-based detection of IG gene rearrangements in cHL whole-tissue specimens, providing a sensitive molecular diagnostic assay for clonality assessment in Hodgkin lymphoma.


FLEX detection system (DAKO) and visualized by EnVision™ FLEX DAB+ Chromogen (DAKO). Epstein-
Barr virus (EBV) status of the HRS cells was determined by in situ hybridization for Epstein-Barr virusencoded small RNAs (EBER) (DAKO). All tissues were scored for EBV status, the presences of HRS cells (CD30 + ), B cells (CD79A + ), T cells (CD2 + or CD3 + ) and the relative fraction HRS cells to total B cells by at least two independent pathologists (Table 1, Supplementary Table S1).

DNA isolation
Clonality analysis was performed on DNA isolated from fresh frozen (FF) and formalin-fixed paraffinembedded (FFPE) whole tumor tissues. DNA extraction from FF tissues was performed on five to ten tissue sections of 10µm with TSE (Tris/Saline/EDTA). In short, tissue sections were lysed overnight at 56°C using TSE buffer (10 mM Tris-HCl pH 7.5 / 0.4 M NaCl / 2 mM EDTA pH 8.0) with SDS (20%) and proteinase K (20 mg/ml), followed by ethanol precipitation and DNA was dissolved in Low TE-buffer (T10E0.1). For FFPE tissues, Chelex DNA extraction followed by column filtering (QIAamp DNA Micro Kit, Qiagen, Hilden, Germany), or the QIAamp DNA FFPE Tissue Kit (Qiagen) was used to isolate DNA from three to six tissue sections of 10µm. DNA concentrations were determined using Qubit (dsDNA BR Assay Kit, Life Technologies, Carlsbad, CA, USA) and DNA quality was assessed using TapeStation (Genomic DNA ScreenTape Assay, Agilent) analysis according to manufacturer's protocol A DIN score of at least 2.5 is considered sufficient DNA quality for further analyses. For samples with a DIN score 2 of around 2.0, the success rate of clonality assessment is less and depends on multiple factors like the percentage of neoplastic cells and the location of the DNA double stranded breaks.

Clonality testing by BIOMED-2/EuroClonality assay
Conventional BIOMED-2 clonality testing was performed on FF and FFPE tissue DNA using BIOMED-2/EuroClonality multiplex PCR master mixes for IGH Tube A, B, C and D and IGK Tube A and B (Invivoscribe Inc., San Diego, CA, USA) or custom-ordered reagents, according to the protocol described by the EuroClonality consortium 1 . Analysis of the obtained PCR products was performed by GeneScanning and results were visualized in GeneMarker® V2. 6.7 (SoftGenetics, State College, PA, USA). Each sample was analyzed in duplicate for both complete and incomplete rearrangements of IGH (IGHV-IGHD-IGHJ, IGHD-IGHJ) and IGK (IGKV-IGKJ, IGKV-KDE, Intron RSS-KDE). For IGHV-IGHD-IGHJ rearrangements, primers for framework (FR) 1, 2 and 3 were used. The molecular conclusions of all gene rearrangement patterns were scored according to the interpretation guideline of the EuroClonality consortium: 'clonality detected (with polyclonal background)', 'oligoclonality/multiple clones detected', 'polyclonality detected', or 'not evaluable' 2 .

Clone identification for cHL cases by IG-NGS
To define threshold values for clone identification in our study that would distinguish a minor cHL clone from a reactive B-cell clone, a series of 30 reactive lymph nodes (study of van den Brand et al. 3 ) was analyzed for the recurrence of identical dominant clonotypes in duplicate analysis for each of the five targets and compared to distribution in cHL tissue. A clonal rearrangement was defined as a clonotype present in duplicate, with an increased abundancy compared to background B-cell clonotypes. The background was determined as the mean percentage of the 5 th , 6 th and 7 th clonotype, since the 1 st to 4 th most abundant clonotypes represent the maximum amount of rearrangements present within one clonal B cell, including bi-allelic IGKV-IGKJ and Intron RSS-KDE rearrangements. The normal distribution of gene rearrangements was examined for each of the targets (IGHV-IGHD-IGHJ (FR3), IGHD-IGHJ, IGKV-IGKJ, IGKV-KDE, Intron RSS-KDE) (Supplementary Figure S1). First, the mean percentage of all overlapping clonotypes in a sample was calculated, followed by ranking from the highest to the lowest value (Supplementary Figure S1 B-C). The top-4 overlapping clonotypes were selected for further analysis by calculating the ratio between the percentage of the clonotype and the background, separately for each sample of the duplicate (Supplementary Figure S1 D). Finally, the mean ratio per sample was calculated. The same procedure was performed for the cHL cases and compared to the results of the reactive lymphoproliferative samples to determine the thresholds for clone identification in cHL samples (Supplementary Figure S1 E). For IG gene rearrangements that involved KDE, slightly different approaches were used. The top-4 overlapping clonotypes for IGKV-KDE were selected based 3 on single-target analysis, after which the ratio's for these clonotypes were calculated based on the combined IGKV-KDE, Intron RSS-KDE analysis. Since Intron RSS-KDE displays only junctional diversity, a different approach was chosen. Here, the two overlapping clonotypes with the highest mean abundancy were selected. The mean percentage of these two clonotypes were used to determine the normal distribution of Intron RSS-KDE rearrangements in reactive lymph node samples. Finally, the values of cHL cases were compared to these results for clone identification. An important note is that there are two Intron RSS-KDE rearrangements, intron -2/0/-3 Kde and intron -0/0/-4 Kde, which are detected with relatively high frequencies in normal B-cell population. A mean percentage that is just above the threshold value for clonality should be interpreted with caution and in the context of clonal rearrangements for other targets.

DNA integrity of the FFPE specimens was assessed using the DEPArray™ FFPE QC Kit (Menarini Silicon
Biosystems, Bologna, Italy). The DNA quality was assessed by the difference in quantification cycles (Cq) obtained with the long amplicon assay (132 bp) and the short amplicon assay (54 bp) for a given sample using the formula: = 1 2 ΔCq . Samples with a QC value of at least 0.3 are of sufficient quality for further analysis. For case 9, 21 single HRS cells, 9 rosetted HRS cells and a leukocyte pool (10 cells) were processed for analysis after isolation from FFPE tissue (two sections of 50 µm) with DEPArray™ sorting technology (Menarini Silicon Biosystems), based on CD30 and PD-L1 expression.
Following HRS cell isolation, genomic DNA was isolated, amplified with Ampli1™ Whole Genome Amplification kit, and DNA libraries were prepared using the Ampli1™ LowPass kit (both Menarini Silicon Biosystems) for sequencing on Illumina MiSeq, as described previously 5

. Copy Number Variation
(CNV) profiles were generated using the CNV caller ichorCNA 6 with default settings. For samples of male patients, this tool re-scales the counts of each bin in Chromosome X with the median counts of this chromosome. Further, unsupervised hierarchical clustering was performed using the R package heatmap3 to analyze the CNV profiles. Supplementary Table S1.  Overview results of BIOMED-2/Genescan analysis for each classic Hodgkin lymphoma sample. Case  Tissue  Type   DIN  score   IGHV-IGHD-IGHJ  (FR3)  IGHD-IGHJ  IGKV-IGKJ  IGKV/  Intron RSS-KDE  Conclusion  IGHV-IGHD-IGHJ  (FR3)  IGHD-IGHJ  IGKV-IGKJ  IGKV-

16
Polyclonal ---* Productivity of the clonotypes is predicted based on an in-frame CDR3 sequence. # Based on the FR3 results the presence of an upstream frameshift resulting in an unproductive B-cell receptor cannot be excluded. Therefore this sample is not marked as a biclonal case in the study. CDR3: complementaritydetermining region 3; aa: amino acid.