Whole-exome sequencing is feasible on a fresh-frozen skin sample of intravascular large B cell lymphoma

Intravascular large B cell lymphoma (IVLBCL) is a rare extranodal B-cell non-Hodgkin lymphoma, accounting for less than 1% of all cutaneous lymphomas with an incidence rate around one case per ten million people per year [1‐3]. The hallmark of this entity is the exclusive growth of neoplastic lymphocytes within the lumina of small-sized vessels. This typical growth pattern has some relevant consequences: the multifaceted clinical presentation, which varies depending on the organs affected; the delay in a timely diagnosis due to misleading clinical suspicion; and the difficulty to retrieve a sufficient number of cells for molecular studies. Any organ may be involved, although lymph nodes are usually spared, whereas central nervous system (CNS) and skin are commonly affected. Indeed, in most cases, there are no lymphadenopathies nor tumor masses at diagnosis [2, 4]. Nonspecific systemic symptoms, such as fever, weight loss and night sweats, may be present. Three main clinical variants are described, with different worldwide distribution and diverse treatment outcomes: cutaneous, classical, hemophagocytic syndrome-associated [5].

Histological examination of tissue biopsy with demonstration of tumor cells is essential for diagnosis. Some sites (such as the CNS) require particularly invasive procedures and/or are encumbered by severe complication risk (such as bleeding for kidney). Random skin biopsy has been increasingly shown as the highest yield approach with a great risk–benefit tradeoff [6‐8]. In fact, its yield is usually higher than that of bone marrow biopsy, even in those patients without clinically apparent cutaneous involvement [6]. Liquid biopsy approaches have recently gained attention, since concentrations of cell-free DNA are exceptionally high in this entity. Nevertheless, no pathognomonic mutations have been found so far, and therefore this tool cannot be used as the only diagnostic method [7‐9].

Given the aggressive behavior of the IVLBCL, deferring diagnosis may lead to multi-organ failure, poor chemotherapy tolerability and fatal complications. Outcomes have been rather improved by increased awareness of the disease (leading to ante-mortem diagnosis) and the addition of the anti-CD20 monoclonal antibody rituximab to CHOP chemotherapy regimen (R-CHOP). Recently, due to the high risk of CNS dissemination at relapse (around 25% at 3 years) [10], the Japanese PRIMEUR-IVL study provided evidence for a new standard of care based on the frontline combination of R-CHOP with a CNS-oriented strategy made of high-dose methotrexate and intrathecal chemotherapy [11]. Nevertheless, around 25–45% [11, 12] patients still relapse, and may die of IVLBCL.

As for nodal subtypes of diffuse large B-cell lymphoma (DLBCL) [13, 14], genomic studies may be of help in the diagnostic workup as well as in prognostic stratification, treatment decisions and follow-up. So far, the low number of neoplastic cells in biopsy specimens and the rarity of this entity have hampered biological studies. To circumvent this problem, in the last years some authors performed targeted next-generation sequencing (NGS) on DNA derived from microdissected tumor cells [7, 15], and the same DNA source has been used for whole-exome sequencing (WES) on post-mortem IVLBCL samples [16], as wells as for WES and RNA-sequencing in the NK/T cell counterpart of this entity [17]. More recently, some groups used cell-free DNA (cfDNA) [7, 9] or mononuclear cells from the bone marrow (BMMNC) [9] to apply WES or smaller targeted gene panels. All the available data point toward an activated B-cell origin of this disease [18], with recurrent mutations in CD79B and MYD88, both occurring in around 50% of cases, frequently in association [7, 9, 15, 16]. However, source material for genomic studies remains a problem: not all cases show bone marrow involvement, cfDNA may be a problematic source of DNA, and tumor cells microdissection is a time-consuming and poorly reproducible technique that may not yield enough DNA for genomic studies.

In this paper, we hypothesized that DNA derived from a fresh-frozen skin biopsy—a routinely used procedure to obtain diagnosis—could be suitable for whole-exome sequencing together with DNA obtained from saliva as germline control.

The study involved the use of human samples, which were collected after written informed consent was obtained. All samples were derived from the same single patient. Samples and data were obtained and managed in accordance with the Declaration of Helsinki. The study was approved by the institutional review board of the Fondazione IRCCS Ospedale Maggiore Policlinico of Milan, Italy, and conducted following the institutional guidelines for retrospective studies.

IVLBCL is a rare entity with poor representation of neoplastic cells in biopsy specimens. Histological analysis of pathological samples is mandatory for diagnosis. Several recent studies showed that random skin biopsy is the approach with the highest diagnostic yield and the lowest side effect and complication rate [6‐8, 26]. Molecular studies are needed to better understand pathogenesis of this disease, to evaluate the prognostic impact of specific genomic alterations, to guide treatment decisions and new treatment development, and to accurately detect early relapse during follow-up.

In consideration of all these elements, we investigated whether skin samples collected during the diagnostic work-up could be used for whole-exome next generation sequencing. Several limitations had to be addressed to get appropriate material for comprehensive genomic studies. First, DNA quality is of paramount importance for subsequent analyses. To this issue, we used fresh-frozen material which is known to allow excellent tissue preservation and to avoid artifacts arising from formalin-fixation. Furthermore, high quality DNA allows to work with limiting DNA quantity. Second, the neoplastic cell fraction is extremely low in tissue biopsies, since tumor cells are located exclusively in small vessels which can’t be extensively and selectively sampled. To solve this problem, we were able to dissect subcutaneous tissue which is enriched in small vessels, and has otherwise low cellularity. This could be performed manually with a scalpel, without the need for laser capture microdissection equipment.

So far, most molecular studies have been limited to targeted-sequencing on DNA derived from either micro-dissected tumor cells or cfDNA [7, 15]. Tumor cell microdissection is an extremely time-consuming and operator-dependent procedure. On the other hand, cfDNA can be problematic too. For example, since no pathognomonic mutations have been found so far, the chance of nonspecific finding is to be considered. As such, MYD88 L265P mutation may be the fortuitous finding of other lymphoproliferative disorders -such as IgM monoclonal gammopathy of unknown significance (MGUS), lymphoplasmocytic lymphoma/Waldenstrom macroglobulinemia, or DLBCL-, rather than the rare IVLCBL. Similarly, CD79B Y196 may be found frequently (around 30%) in activated-B cell (ABC)-DLBCL, especially in those aggressive cases with extranodal clinical presentation, the so called “MCD” or “cluster 5” subgroups, as defined by Schmitz and Chapuy, respectively [13, 14]. Conversely, in our approach, these mutations can be directly assigned to the tumor clone present in the biopsy.

Targeted studies focused on a panel of genes, selected on the assumption of a non-GCB origin of IVLBCL, have been important in confirming aberrations in pathways such as the B-cell receptor (BCR)-signaling pathway, the Toll-like receptor/Interleukin-1 receptor (TLR/IL-1R) pathway, and the nuclear factor kappa B (NFκB) pathway. However, more appropriate and specific gene panels could be established only after comprehensive genome-wide studies. In fact, to date there are no genomic data able to explain the typical histological pattern of this disease nor to confirm mutations in genes whose products were previously described as potential players of intravascular distribution (such as CD29 or CD54) [5].

Until recently, the only attempt of comprehensive genomic studies was performed on DNA derived from both cfDNA and BMMNC, although the cohort was entirely formed of Japanese patients, and therefore may not be representative of the worldwide IVLBCL cohort. Yet no genomic features leading to the typical intravascular pattern have been found. Furthermore, analysis of bone marrow samples is limited to the cases with bone marrow involvement. More recently, a WES study was performed by Kodgule and colleagues on post-mortem tissues involved by IVLBCL [16]. In their study, they found recurrent mutations in the RAC2 gene, and postulated that this mutation may play a role in IVLBCL cells adhesive features. However, in the aforementioned WES study by Shimada and colleagues, RAC2 mutations were identified in only a minority of cases, and this particular finding was not emphasized [9]. We were not able to detect any RAC2 mutations despite thorough evaluation. Given that our analysis was confined to a single patient, and the two other studies reported RAC2 mutations in only a subset of patients, this finding underlines the possibility that, for some IVLBCL cases, exclusive intravascular location may be attributed to different mediators [9]. Further comprehensive studies are needed to evaluate the recurrence and significance of RAC2 mutations in IVLBCL.

In this study, we found that fresh-frozen subcutaneous sample allows comprehensive genomic studies. In fact, DNA quality and quantity were high, and the cancer cell fraction could be artificially enhanced by selection of the subcutaneous selection. The resulting tumor DNA fraction, albeit low, was amenable to sequencing analysis at medium depth. However, it was likely still too low to reliably identify CNAs. Confirming the feasibility and robustness of our approach, our results align with those obtained from workflows using different sequencing techniques, sample types and manipulations, and geographical areas. In fact, we confirmed the two most frequently described mutations in IVLBCL, CD79B and MYD88 [7, 9, 15, 16]. Moreover, we found a similar number of mutations as compared to the two other WES analyses [9, 16]. In line with other reports, we confirmed that our sample bears multiple mutations in genes targeted by SHM (PIM1, TMSB4X, MPEG1 and OSBPL10) [9, 16]. Since we could only analyze one sample, we cannot exclude that genes with multiple mutations in our dataset could also be artefactual or arise from other localized hypermutational processes such as those mediated by aberrant activity of apolipoprotein B mRNA editing enzyme catalytic polypeptide (APOBEC) [27]. Although limited by the small sample size which is partly due to the extreme rarity of this entity, we believe our study may represent an important proof-of-concept which could be further explored. Our methodology, which is time-efficient and cheap, can be adopted by any pathology laboratory, and could be easily integrated in the standard diagnostic workup of the patient, which is based on random skin biopsy even in the absence of skin lesions. The main limitations of our approach include its operator-dependent nature during the cutting of the biopsy specimen and the necessity to freeze the sample which can be resource-intensive. On the other hand, identifying subcutaneous tissue before cutting is easily feasible, and freezing IVLBCL biopsies should not strain resources, given its rarity.

Therefore, this approach, if validated, may be employed in future studies to better define the genomic landscape of this rare entity, through whole genome sequencing studies which can be carried out at increasing depths using modern technologies and can detect subclonal populations [28]. Extensive and widespread adoption of such pipeline may lead to recognize the specific aberrations associated with this entity, its different clinical presentations, response to treatment, and CNS relapse risk. In case pathognomonic mutations should be found, they could also be used as an important tool to confirm diagnosis and guide measurable residual disease assessments in follow-up. Unravelling the characteristic pathogenetic features may in turn lead to personalized treatments through the use of targeted panels [29‐31]. Interestingly, MYD88 L265P has already been described as a target of ibrutinib, whereas little is known about CD79 mutations and polatuzumab vedotin efficacy. Moreover, since the microenvironment contribution on disease pathogenesis is likely limited, IVLBCL may be a good model to identify tumor-intrinsic key drivers, as opposed to DLBCL which is much more influenced by interaction with non-neoplastic cells most times.

In conclusion, fresh freezing a portion of the random skin biopsies performed during the diagnostic work-up and restricting analysis to subcutaneous tissue results is an easily reproducible and cost-effective approach that may be useful for subsequent comprehensive genomic analysis which are still strongly needed in this entity. To the best of our knowledge, this is the first study to use fresh frozen tissue IVLBCL sample collected from living patient for whole-exome sequencing.