XVII. Treatment of advanced-stage Hodgkin lymphoma

Advanced-stage Hodgkin lymphoma (HL) usually includes all patients diagnosed in Ann Arbor stages III and IV. Many groups also include patients with stage IIB and additional risk factors such as large mediastinal mass and/or extranodal disease. Historically, less than 5% of these patients survived when left untreated or received single-agent chemotherapy. With the development of multi-agent chemotherapy such as MOPP (mechlorethamine, vincristine, procarbazine and prednisone) or ABVD (doxorubicin, bleomycin, vinblastine and dacarbazine), the disease became curable, and a multicenter trial demonstrated that ABVD was better than MOPP in terms of freedom from progression (80.8% vs 62.8%; p < 0.002) and overall survival (77.4% vs 76.9%; p = 0.03) [1]. These findings were confirmed in different trials with longer follow-up.

The next generation of clinical trials included hybrid regimens such as MOPP/ABVD, cyclophosphamide, oncovin, procarbazine and prednisone (COPP)/ABVD or MOPP/ABV or regimens containing even more drugs in rapidly alternating sequence (Table 1). Examples were MOPP/epidoxirubicin, bleomycin, and vinblastine (EBV)/cyclophosphamide, adriamycin, dexamethason (CAD) or hybrid regimens such as chlorambucil, procarbazine, prednisolone, vinblastine, doxorubicin, vincristine, bleomycin, and etoposide (ChlVPP/PABlOE) or chlorambucil, vinblastine, procarbazine, and prednisolone/etoposide, vincristine, and doxorubicin (ChlVPP/EVA) [2]. An alternative shorter US regimen, Stanford V, gave challenging results in a phase II trial. With 142 patients treated at a single centre, the 5-year tumour control was 89% and OS 96%. Importantly, 91% of patients in this trial received additional radiotherapy. A small prospectively randomised multicenter trial compared ABVD with Stanford V and MOPP-EBV-CAD demonstrating that Stanford V was associated with poorer failure free survival (FFS) (67% vs 83% or 85%) when compared with ABVD or MOPP-EBV-CAD [3]. In addition, a larger intergroup trial compared also Stanford V with ABVD [4]. In this trial, the complete remission rates were rather similar (73% for ABVD and 69% for Stanford V), and there was no difference in FFS at 5 years (74% for ABVD and 71% for Stanford V). More patients received additional radiotherapy with Stanford V (71% vs 40%), and this regimen was more toxic. Thus, the challenging data initially reported for Stanford V in single centre phase II trials were not reproduced in the multicenter setting.

Based on statistical modelling, the German Hodgkin Study Group (GHSG) developed the BEACOPP regimen (bleomycin, etoposid, doxorubicin, cyclophosphamide, vincristin, procarbazine and prednisone). After initial dose finding studies, two variants, BEACOPP_baseline and BEACOPP_escalated, were compared to the standard chemotherapy at that time, eight cycles of COPP alternating with ABVD. In the subsequent multicenter phase III trial, 1195 patients with advanced-stage HL were randomised. Overall, there was a clear superiority of BEACOPP_escalated over BEACOPP_baseline and COPP/ABVD at 5 years [5]. The results were more obvious at 10 years follow-up with significant differences between BEACOPP_escalated and COPP/ABVD in terms of tumour control (18%) and OS (11%) [6]. Importantly, the impact of BEACOPP_escalated was most obvious in the intermediate prognostic score (IPS 2–3) representing the largest subset of advanced-stage patients.

However, eight cycles of BEACOPP_escalated were associated with more toxicity including leukocytopenia and thrombocytopenia, anaemia, infections, fertility and secondary leukemias. The GHSG follow-up studies Hodgkin's Disease (HD12) and HD15 consequently aimed at reducing toxicity albeit maintaining efficacy. In the HD15 trial, about 400 centres from five European countries contributed 2196 patients. At 5 years, six cycles of BEACOPP_escalated were significantly better than the old standard of eight cycles, both in progression-free survival (90,3% vs 85,6%) and overall survival (95.3% vs 91.9%) [7]. In addition, the treatment-related mortality was only 0.8% with six cycles as compared with 2.1% with eight cycles. Also, the number of secondary leukemias was much lower (0.3% vs 2.7%). The better tolerability might in part relate to fewer patients receiving additional radiotherapy (11% in HD15 as compared with >70% in HD9). Thus, six cycles of BEACOPP_escalated are the current standard of care for HL patients in advanced stages for the GHSG.

Comparing BEACOPP with ABVD, the challenging results associated with BEACOPP_escalated in advanced-stage HL lead to intensive discussions on the standard of care for advanced-stage HL. More patients survive after first-line treatment with BEACOPP_escalated, but there is more toxicity. In contrast, ABVD is better tolerated but cures fewer patients. Can the higher relapse rate with ABVD be compensated by effective treatment at relapse?

Another argument against BEACOPP_escalated becoming the undisputed standard of care was the lack of direct comparisons because the GHSG trials had initially compared BEACOPP_escalated with COPP/ABVD. A total of four smaller trials subsequently compared ABVD with BEACOPP in different versions. Taken together, a total of 1222 patients were included in these four trials. The results at 5 years indeed show significant superiority for those patients treated with BEACOPP (difference ranging from 12% to 18%). The 5-year overall survival difference ranged between 4% and 8% without significance. Because all trials were powered for differences in tumour control but not for overall survival, these data clearly support the superiority of BEACOPP_escalated over ABVD. In addition, a more recent network analysis confirmed these findings with more than 10 000 patients and 47 033 patient years of follow-up showing a reconstructed individual survival advantage of 10% for 6 cycles of BEACOPP_escalated over ABVD [8].

Current trials

With the advent of positron emission tomography (PET), this technique has not only improved the staging accuracy for HL patients but also indicated prognostic significance. A number of phase II trials thus currently use PET in order to identify good or poor responders early so that treatment can be adapted accordingly. Most groups start with two cycles of ABVD and continue this treatment in those who are PET negative; PET positives are then typically switched to BEACOPP_escalated. The alternative is to start with BEACOPP_escalated and de-escalate to ABVD in PET negative cases after two cycles of chemotherapy. In the GHSG HD15 trial, PET was used to assess the activity of residual masses ≥2.5 cm after chemotherapy so that only a minority of 11% PET+ patients (n = 191) received additional radiotherapy. The negative predictive value was 94.1% at 12 months. At 4 years, progression-free survival for PET− was 92.6% and 86.2% for PET+ patients suggesting that this pragmatic approach is feasible, at least in those treated with BEACOPP_escalated.

Future treatment of HL patients in advanced stages will be more adapted to response and risk profile. In addition, smarter combinations of both, ABVD or BEACOPP, with new drugs such as the anti-CD30 anti-drug conjugate brentuximab vedotin are currently being evaluated in prospective randomised trials. In addition, immune checkpoint inhibitors will very likely also find a role in first line so that the treatment for HL in advanced stages will become more specific and less toxic.

Conflict of interest

Prof. Dr Andreas Engert reports Millennium/Takeda, Affimed and Bristol-Myers Squibb.