Ferritin, sex, and the wildfire of inflammation in Hodgkin lymphoma

Classical Hodgkin lymphoma (cHL) is clinically stratified into early favorable, early unfavorable, and advanced disease. The classification, which largely relies on radiographic staging, impacts prognosis and treatment, including length and type of chemotherapy or applicability of radiation therapy. However, features related to systemic effects of cHL, like presence of B-symptoms or elevated erythrocyte sedimentation rate (ESR), have long been recognized to also portend worse prognosis, particularly in early-stage disease. In advanced-stage cHL the risk of treatment failure is typically described using the International Prognostic Score (IPS) which was derived from experience of 4695 patients treated principally in the 1980’s.[1] As treatments changed and patient outcomes improved, the prognostic accuracy of IPS has decreased.

Our understanding of cHL has markedly advanced within the past 15 years. The disease is driven by immune evasion of the malignant Hodgkin or Reed-Sternberg (H/RS) cells (which are germinal center B-cells crippled by loss of B-cell receptor and destined for elimination) from the surrounding inflammatory infiltrate that constitutes over 98% of the tumor mass. Histology of cHL depends on the specific composition of this infiltrate and the extent of cytokine-induced fibrosis. Increasingly sophisticated studies, incorporating laser-capture microdissection of H/RS cells, immunohistochemistry, fluorescent in situ hybridization, and culminating in single-cell DNA sequencing, have revealed the molecular basis of the ability of H/RS cells to escape immune surveillance and to overdrive inflammation. These cells nearly universally harbor copy gains of chromosome 9p24.1 which results in overexpression of the programmed cell death 1 (PD-1) ligands and activation of pro-inflammatory JAK-STAT pathway.[2] Other mechanisms like loss of the major histocompatibility complex (MHC) antigens or constitutive activation of the NF-κB signaling pathway further enable H/RS cells to manipulate the local microenvironment towards T-cell exhaustion, recruitment of myeloid-derived suppressor cells, and unfavorable polarization of the macrophage population. Insights into the pathobiology have been promptly translated into treatment of cHL with a remarkable success of immune checkpoint inhibitors in relapsed/refractory disease.[3] As the treatment of cHL evolves into personalized strategies that incorporate first-line immunotherapy, the lack of biology-driven biomarkers that could enhance prognostic stratification has become a notable shortcoming. In the United States, while the strategy of initial doxorubicin, bleomycin, vinblastine, and dacarbazine (ABVD) chemotherapy modified by the interim positron emission tomography (PET) remains in clinical use, it is no longer a reference arm for future research. Instead, the current intergroup S1826 phase 3 trial for advanced-stage cHL (NCT03907488) uses AVD in combination with either brentuximab vedotin or nivolumab.

In this issue of Leukemia & Lymphoma, Karakatsanis et al. present an in-depth evaluation of ferritin—an easily available marker of systemic inflammation—in a cohort of 529 patients with cHL treated almost exclusively with ABVD chemotherapy.[4] Similar to a few prior studies, the authors show an independent prognostic value of ferritin for progression-free survival (PFS) though not for OS, and lacking power to evaluate the subgroup with early-stage lymphoma. Critically, they demonstrate that ferritin is sensitive to sex-dependent iron stores and thus different cutoffs for men and women performed better than a uniform cutoff for prognostication. While other markers like ESR, C-reactive protein, or hypoalbuminemia correlate with the pro-inflammatory effects of cHL, ferritin and soluble interleukin-2 receptor (SIL2R) may best reflect the T-cell and macrophage activation that are mechanistically involved in the failed immune response against the H/RS cells. However, ferritin, apart from being sensitive to iron stores, is also affected by disorders of iron metabolism (thalassemia, hemochromatosis) and by liver disease, particularly non-alcoholic steatohepatosis (NASH) which is common in industrialized countries. In the present study, high ferritin correlated with age and high body mass index, possibly reflecting its non-specific association with age- and obesity-related inflammation. Ferritin also correlated (in various directions) with almost all other studied laboratory parameters, ranging from platelet count to serum beta-2 microglobulin, thus obscuring its interpretation.

The study by Karakatsanis et al, is limited by its retrospective design and lack of external validation but raises interesting questions to inspire further research on prognostic and predictive biomarkers for cHL in the immunotherapy era. Some of these questions are:

• What is the relative value of biomarkers derived from serum, the cHL tumor itself (and in that case, H/RS, inflammatory cells, or both), and germline polymorphisms related to host immunity?

• Which biomarkers provide insight into the biology of cHL, and which simply correlate with the amount of disease?

• Which biomarkers are best measured prior to treatment, and which provide more information after the start of therapy?

• Can immune-oriented biomarkers show prognostic relevance in early stage cHL?

• Is their relevance retained with the checkpoint inhibitor or brentuximab vedotin-containing therapy? Might they have a predictive value for checkpoint inhibitors, considering that brentuximab vedotin primarily serves as a targeted chemotherapy delivery system?

Serum biomarkers like ferritin offer the advantage of easy measurement over biopsy or PET-derived assessments, but research on their use in cHL has not provided a consistent picture. In the study from the French Goupe d’etude des Lymphomes de l’Adulte, baseline IL-1RA, IL-6 and sCD30 were all associated with event-free survival.[5] In contrast, an analysis from the German Hodgkin Study Group found only IL-10 to predict early treatment failure.[6] In the S0816 trial of response-adapted therapy for advanced cHL, baseline or early on-treatment levels of IL-10, thymus and activation-related chemokine (TARC), macrophage-derived chemokine, or soluble CD163 turned out to be not informative, but their end-of-treatment levels predicted subsequent PFS.[7] In the Mayo study of 140 patients, pre-treatment IL-2 receptor and IL-6 (which correlated also with TARC and soluble CD30) were the only (and highly) prognostic cytokines.[8] These discrepancies underscore the necessity of thorough validation of serum biomarkers, ideally in the context of prospective data—though not necessarily limited to clinical trials, which may be affected by selection bias. Molecular assays might prove more specific, yet they remain costly and challenging to validate as well. For example, despite initial promise, gene expression profiling has not shown consistent utility for risk stratification of cHL. Measurement of circulating tumor DNA (ct-DNA) has emerged as a tool for genotyping of cHL and for quantification of tumor burden on therapy, though it remains to provide prognostic information from baseline measurement.[9] Recently, an approach evaluating T-cell repertoire diversity, clonal expansion and specific subtypes of effector cells showed correlation with response to checkpoint inhibitor therapy in cHL.[10]

As the novel immunotherapeutic agents are gradually incorporated into first-line therapy, it will be critical to examine both molecular and serum biomarkers in clinical trials for maximal rigor. The most impactful might be biomarkers that can identify excellent responders to immunotherapy alone, who might be considered for phase-out of chemotherapy in a subsequent iteration of trials. Delineating a low-risk group who could be cured with a de-escalated RATHL strategy without the toxicities of brentuximab vedotin-AVD combination would also be of significant practical value. If a serum marker as simple as ferritin, carefully adjusted to sex-specific reference ranges, could contribute useful information in the context of new therapies, it would be an unexpected, yet welcome advance.