Separating the wheat from the chaff in cHL

In this issue of Blood, Reichel and colleagues¹  combine two techniques—flow sorting and low-input sequencing—to show that the “elusive Reed-Sternberg cell”²  of classical Hodgkin lymphoma (cHL) can be captured and interrogated exomewide.

Neither low-input sequencing nor flow sorting of Hodgkin and Reed-Sternberg (HRS) cells is by itself a new technique,^3,4  but their combination in the context of Hodgkin lymphoma—as is often the case with the application of existing techniques in new areas—provides at once a proof-of-principle template for future investigation, as well as new data that advance knowledge in the field immediately.

Exome sequencing of Hodgkin lymphoma cell lines has already revealed a plethora of mutations,⁵  including some of those outlined in this article. Cell lines are not primary tumors, however. A handful of cell lines is unlikely to represent the diversity of mutations present in patients’ tumors, and the gene expression programs of cell lines differ somewhat from those of tumors in situ.⁶  Other investigators have isolated and studied HRS cells by microdissection,⁶  a time-consuming procedure that limits the total number of cells that could be analyzed and the total amount of DNA or RNA that could be produced for downstream analysis. Reichel et al aimed to overcome these limitations with a simple but powerful idea. Tumors were flow sorted according to a variety of surface markers,⁴  and thousands of viable HRS cells were separated from the surrounding lymphocytic milieu. DNA was amplified and subjected to exome sequencing for analysis of copy number changes and somatic mutations.

The authors identified many recurrent (defined in this study as occurring in 2 or more cases) copy number alterations and somatic mutations, but a few stand out. Mutations in β-2 microglobulin (B2M) are notable because the association of major histocompatibility complex class I (MHC-I) loss and Epstein-Barr virus–negative Hodgkin lymphoma has been known for many years;⁷  the present study reveals that the mechanism in primary tumors is truncating mutations of B2M, and extends this result by expressing B2M in Hodgkin lymphoma cell lines, restoring MHC-I expression. In a validation cohort (N = 145), B2M loss was prognostic, but not in a multivariate analysis, owing to the strong association of B2M loss with younger age. TNFAIP3, a tumor suppressor and inhibitor of nuclear factor κB (NF-κB), has previously been shown to be mutated in Hodgkin lymphoma cell lines and microdissected Reed-Sternberg cells.⁸  Its strong showing in the present study validates the role of NF-κB in Hodgkin lymphoma and lends confidence to the veracity of other reported novel mutations.

The authors speculate that these discoveries may lay the groundwork for personalized medicine based on a specific genetic profile in cHL. This scenario may one day come to pass, but flow sorting a minute population from background, then amplifying tiny amounts of DNA before ultimately sequencing a potentially broad range of mutations is a somewhat complicated workflow from a clinical laboratory perspective. One possible alternative that leverages the valuable data resulting from these studies might instead be using high-sensitivity polymerase chain reaction techniques to look for now-known recurrent mutations from the bulk nucleic acid collected from a lymph node biopsy.

Still, this study has clinical implications. Alterations in B2M and TNFAIP3 segregate strongly with subtype; using molecular profiles to determine the histologic subtype of cHL may allow more objective disease classification and clinical trial stratification. Retrospective evaluation of specific alterations may uncover new prognostic and predictive markers to guide even nontargeted therapy. Commendably, many of the detected recurrent alterations are paired with drugs in Table 3 of the article.

Although this article provides an important template for future studies, it is far from definitive. As the authors acknowledge, a low number of samples (N = 10) and a modest sequencing depth (median 48×) mean that less-prevalent (in terms of both the population of cHL patients and the population of cells within an individual patient) mutations may be missed. Other investigators have recently shown a high proportion of PTPN1 mutations in Hodgkin lymphoma primary samples and cell lines,⁹  whereas the present article does not report this mutation as recurrent (although PTPN1 is deleted in one case and mutated in another according to the data supplement). This finding suggests that additional genetic alterations still lay undiscovered.

The estimated average tumor purity in this study was 75%, which, although low by modern leukemia and lymphoma sequencing standards, is an improvement in signal-to-noise ratio for Hodgkin lymphoma given the incredible rarity of the Reed-Sternberg cell in its background nest of inflammatory infiltrate. Previous studies relying on laser capture microdissection could also increase the signal-to-noise ratio, but at the cost of time, effort, and still-low cell numbers.

This work raises additional questions. Variant allele frequencies range from 2% to 100% within single cases. Does this imply intratumoral heterogeneity? Do individual or multiple HRS cells exist as clones and subclones with different alterations and characteristics within the same lymphoma, and if so, what would that mean for molecularly targeted therapy?

Overall, this work is exciting for several reasons. It advances our knowledge of the genetic makeup of cHL but also demonstrates the feasibility of a new technique that could in the future be extended to provide additional information about transcription and potentially protein expression. The demonstration that restoration of B2M loss can reverse a characteristic phenotype is an important functional validation of a key finding and is timely and relevant because immune therapies are now being explored in Hodgkin lymphoma.¹⁰