1.1. Classical Hodgkin’s Lymphoma (cHL) Particularities

Classical Hodgkin’s lymphoma (cHL) is a rare and curable malignancy with an annual incidence representing 10% of new lymphomas and a prevalence of 1% of all cancers in Western countries [1]. Patient outcomes are mostly excellent with multiagent chemotherapy (escalated BEACOPP, ABVD) and modern radiation techniques with a five-year progression-free survival (PFS) rate of 65% to 90% according to advanced versus localized stage disease and standard clinical risk factors ^[2][3][2,3]. Patients achieving an early metabolic rapid response after two cycles of frontline chemotherapy display excellent outcome and are proposed for treatment intensity de-escalation after interim positron emission tomography (PET) assessment [4]. However, approximately one-quarter of patients will display progressive disease or relapse, pinpointing the urgent need to determine the underlying biological processes involved and to select useful biomarkers. At present, no pathognomonic biomarker exists for cHL. To date, the genetic landscape of cHL has been incompletely described because Hodgkin and Reed-Sternberg (HRS) tumor cells are very scarce (0.1–3% of cells in the tissue) [5], hampering molecular biology analyses with techniques that lack sensitivity.

Standard lymph node removal or tissue biopsy is the recommended procedure for the lymphoma diagnosis and is the routine method for tumor genetic profiling, but this invasive method is associated with numerous caveats: hemorrhage, infections, anesthetic risks, or poor-quality fine-needle biopsy with artifacts or sampling issues [6]. In addition, the rarity of HRS cells augments pathological diagnosis complexity in small biopsy samples. In line with these tumor DNA access difficulties, several teams hypothesized that discovering tumor specific somatic alterations would be more appropriate in the bloodstream, i.e., in cell-free DNA (cfDNA) extracted from plasma. Indeed, various proof-of-concept studies have recently shown that cHL molecular analysis is feasible using cfDNA and highly sensitive methods ^{[7][8][9][10][11]}[7,8,9,10,11]. Despite the scarcity of HRS cells and a typical lower tumor volume, many studies have indicated that success rates for cfDNA detection in cHL are close to those achieved for diffuse large B cell lymphoma (DLBCL). This finding suggests that cHL seems to display a higher trend to release cfDNA than DLBCL. This finding may be related to an increased fraction of tumor cells in apoptosis but also related to the nature of the alterations in the nuclear DNA of reed Sternberg cells [12].

1.2. Cell-Free DNA Physiopathology

cfDNA was discovered in the bloodstream several years ago and was first assessed in solid tumors. Previous interesting works in solid tumors settings established that cfDNA fragments may invade distant cells in other tissues, modifying the biology of these cells and contributing to the onset of metastases ^[13][14][13,14]. For cHL, no similar data exist. cfDNA is predominantly freed in plasma by cells in apoptosis and is in addition actively secreted by certain tumor cells or liberated during necrosis phenomenon (see Figure 1). These facts are established by the existing broad spectrum of cfDNA fragment sizes (from 0.150 to several kilobases) and the demonstration of similar genetic profiles in tumor genomic DNA (gDNA) and cfDNA ^[15][16][17][15,16,17]. Of note, cfDNA exist in healthy subjects and mainly derived from apoptosis of hematopoietic cells. This normal cfDNA is typically detectable in very small quantities in blood [18]. However, these levels increase 15-fold 30 min postexercise and return to normal levels thereafter [19]. The physiological role of cfDNA remains undoubtedly elusive, but a twofold increase in cfDNA levels after the psychosocial stress test and five-fold increase after exhaustive treadmill exercise were recently demonstrated, suggesting that cfDNA is a biomarker of molecular stress [20]. Biological and environmental variables may also modulate cfDNA release, including sex, body composition, age, smoking, exercise, autoimmune disorders, comorbidities, infectious disease, inflammatory conditions, oxidative stress, and pregnancy ^[21][22][21,22]. In addition, cfDNA clearance is also complex and may imply DNase I activity [23], renal clearance [24], and uptake by the liver and spleen followed by macrophage elimination [25]. Several teams measured that the cfDNA half-life in blood is comprised in a range of 16 min to 2.5 h [26] but is largely dependent on several patient settings: healthy subjects vs. cancer patients, before surgery/chemotherapy/radiotherapy or after, and at rest vs. after physical activity [22]. Furthermore, cfDNA clearance is also influenced by binding to cell-surface receptors [27] and several serum proteins [28] (albumin, fibrinogen, prothrombin, and C-reactive protein), the levels of which may considerably vary during the cancer course. All these features complicate biological studies on this topic. It is important to note that cfDNA combined both “normal/non-tumoral” cfDNA and circulating tumor DNA (ctDNA) fragments ^[18][29][18,29], and there is no tool to separate cfDNA arisen from cancer cells and cfDNA liberated by normal cells, which begs the question of background noise and sensitivity of the technologies used in liquid biopsy works to detect somatic variants. Of note, several research teams in solid tumors detected cfDNA in other human fluids, such as cerebrospinal fluid (CSF) in primary central nervous system lymphomas (PCNSLs) with MYD88 L265P mutation assessment ^[30][31][30,31], urine [32] in the bladder cancer, sputum [33] in lung cancer patients, and uterine lavage fluid in patients with endometrial cancers [34]. In addition, stool DNA may also be valuable in patients with colorectal carcinoma with improved detection rates and a commercially available tool (Cologuard™ assay) [35]. However, these sources of cfDNA seems irrelevant for cHL with no published data to date and no possibility to determine their value for disease burden assessment or genotyping at the time of diagnosis.

Figure 1. Schematic overview of cell-free DNA in classical Hodgkin lymphoma. Abbreviations: MRD: minimal residual disease.

Finally, relevant advantages of plasma cfDNA testing in cHL include: (i) simple venous puncture to obtain sample, (ii) measurable tool which may be performed at any time during patient’s journey (iii) dynamic assessment of clonal evolution, and (iv) less spatial heterogeneity than tissue biopsy genotyping ^[36][37][38][36,37,38].

1.3. Cell-Free DNA Molecular Tools

We know that the optimal way for noninvasive liquid biopsy testing is to extract cfDNA from plasma after blood puncture with nucleic acid preservation tubes (for example, Roche or Streck cfDNA BCT ^® [39]). These tubes should then be promptly processed (within 6 h after venous puncture) with consecutive low- and high-speed [40] centrifugations to reduce leukocyte lysis. The intensity of low temperature storage room (−20 °C or −80 °C) remains controversial, but the relevance of leukocyte stabilization tubes is clearly established for easier use [41], especially in multicentric studies. Research teams may experience altered sample quality if they do not satisfy the optimal preanalytical requirements [42]. cfDNA extraction from plasma samples is easy and feasible in most academic laboratories using commercial kits ^[43][44][43,44]. As mentioned above, in cHL, cfDNA comes from rare lymphoma cells and normal cells, thereby necessitating highly sensitive methods for accurate measurement of somatic alterations.

It was previously established that tumor cells in cHL arise from B-cells [45], and clono-specific B-cells are detectable in cHL patients’ blood samples [46]. Normal C-cells and lymphoma cells both expressed B-cell receptor (BCR). BCR variety is a consequence of variable-diversity-joining (VDJ) genes rearrangement during lymphopoiesis. This mechanism provides specific clonotypes, and so each tumor-specific VDJ profile may be considered as a “barcode” for noninvasive tracking of lymphoma in liquid biopsy. Indeed, using ClonoSEQ technology (Adaptive Biotechnologies, Seattle, WA, USA) Oki et al. described a small proof of concept series of seventeen patients, of whom eleven harbored a detectable lymphoma-specific clonotype in tumor biopsies, 8/11 (73%) displayed the same clonotypes in plasma cfDNA, and 33% exhibited the same clonotypes in PBMCs [47]. Using universal VDJ and IdK primers instead of tumor-specific primers, it is possible to detect clonospecific sequences in a single cfDNA sample. This may grant to disclose exhaustive patients’ immunoglobulin repertoire and monitor individual subclones. This ClonoSEQ assay is the sole FDA-cleared minimal residual disease (MRD) tool in lymphomas. Nevertheless, to our knowledge, ClonoSEQ technology results in cHL patients have not been reproduced by other teams, and the sensitivity and specificity of this technique in cHL remain unclear at the moment. Moreover, this technique requires the initial tumor biopsy material for exact assessment of the VDJ profile before tracking it in the blood. In addition, VDJ rearrangements may be unproductive or abortive [48], so the ClonoSEQ method may not work for in this situation, restricting the informativeness of this tool.

In contrast, next-generation sequencing (NGS) gene panel tests may detect concordant potentially “actionable” somatic mutations in the patients’ plasma (cfDNA) and biopsy (gDNA) of lymphoma patients [49] and may contribute to decide appropriate salvage treatments in relapsed/refractory aggressive B-cell lymphoma using new target therapies currently in development. For example, CAncer Personalized Profiling by deep Sequencing (CAPP-seq) is a powerful method for cfDNA measurement that allows deep DNA sequencing and grants an easy detection and quantification of ultralow abundance genetic alterations [50]. CAPP-seq relevance was well described in non-Hodgkin and Hodgkin lymphoma patients. This technology is able to measure disease burden, detect early relapse before radiological progression, perform cell of origin (COO) classification, separate indolent follicular lymphomas and those at risk for high grade transformation, and monitor variants’ clearance in chemosensitive patients versus non-responders patients who display persistent genetic alterations in plasma after treatment ^{[8][51][52][53]}[8,51,52,53]. However, such results are only possible at high cost given the elevated number of genes included in the panels and in trained research teams with experienced bioinformaticians able to combine barcoding and unique molecular identifiers (UMIs) with integrated digital error suppression [54]. To date, CAPP-seq is not commercially available. Another report recently assessed cfDNA using real-time PCR in an impressive cohort of 155 pediatric cHL. In this work, the authors showed that baseline cfDNA level is higher in cHL than in healthy subjects, and that higher cfDNA concentration is linked to B-symptoms and inflammatory syndrome. The authors also established that the augmentation of cfDNA concentration after one cycle of chemotherapy led to unfavorable outcome [55].

Finally, digital PCR (dPCR) is a quick, simple and barely costless tool which only needs a small amount of plasma cfDNA. dPCR process dilutes and partitions DNA samples into thousands microcompartments (i.e., microscopic PCR reactors) with each one including a single copy or no copies of the target region [56]. It is then easy to quantify the exact normal or mutated DNA copies number by counting the number of positive compartments with the fluorescent probe corresponding to the wild-type or mutated region. The advantages of dPCR include rapid implementation, the lack of a need for a bioinformatics pipeline and high sensitivity (10⁻⁵ detection limit), making it a relevant tool when used independently and in addition to NGS for hotspot single-nucleotide variant (SNV) detection, such as XPO1 E571K (primary mediastinal B-cell lymphoma and cHL) [44]. Indeed, dPCR is based on single-point mutations quantification, and so is not designed to provide the complete molecular landscape of the patient’s lymphoma and, therefore, probably not suitable for molecular response assessment given frequent subclonal evolution. In addition, treatment sensitivity could remove those subclones from cfDNA profiles, so other clones could arise or survive with a false negative dPCR assay. A solution could be to multiplexe dPCR assays to test several hotspot mutations in the same experience, and indeed new dPCR tools can do so (including RainDance^® or Biorad^®) ^{[57][58][59][60]}[57,58,59,60]. Furthermore, low amount of cfDNA in some plasma samples impairs the capability to measure molecular response given insufficient haploid genome equivalent quantities. Of note, false positives/background noise issues and technical limits of variant detection are still debated ^[61][62][61,62]. In addition, to date, no published multicenter study of dPCR MRD approaches exists in cHL patients.