Impact of risk factors on outcomes in early-stage Hodgkin's lymphoma: an analysis of international staging definitions

Background: In early-stage Hodgkin's lymphoma (HL), treatment according to the early favorable or unfavorable subgroup is guided by staging definitions, which differ between various study groups worldwide. We analyzed risk factors used in different international staging systems and their impact on the outcome of early-stage HL patients.

Patients and methods: In 1173 early-stage HL patients treated homogenously within the German Hodgkin Study Group (GHSG) trials HD10 and HD11, the impact of three staging systems developed and used by the GHSG, the European Organization for Research and Treatment of Cancer (EORTC), and the National Comprehensive Cancer Network (NCCN) in discriminating risk groups for progression-free survival (PFS) and overall survival (OS) was assessed and the relevance of their single risk factors was investigated.

Results: All the three staging systems defined an unfavorable risk group out of early-stage patients of comparable size (56%, 55%, and 57%), having a significantly poorer PFS and OS as compared with the corresponding favorable group; 5-year differences between early favorable and early unfavorable in terms of PFS were 9.4% (HR 2.61, 95% CI 1.74-3.91), 6.7% (HR 2.10, 95% CI 1.41-3.13), and 8.6% (HR 2.14, 95% CI 1.45-3.16) with the GHSG, EORTC, and NCCN definition, respectively. Sensitivity was high for all systems (84%, 79%, and 83%); however, there was a low specificity with high rates of false-positive results (1-specificity 54%, 53%, and 55%, respectively). Models of high sensitivity included risk factors associated with large tumor burden and high tumor activity. Most risk factors for tumor-specific end points were also predictive of OS.

Conclusions: Differentiating between a favorable and an unfavorable risk group has significant impact on PFS and OS in early-stage HL patients in the modern treatment era. Risk-adapted treatment strategies using new risk factors with higher specificity are needed.