Newly diagnosed advanced-stage Hodgkin lymphoma (HL) has become a curable malignancy in the vast majority of patients.1, 2 With response-adapted treatment comprising four or six cycles of bleomycin, etoposide, doxorubicin, cyclophosphamide, vincristine, procarbazine, and prednisone (BEACOPP) in escalated dose (eBEACOPP), outstanding cure rates are achieved through a comparatively short and safe treatment approach.3, 4

Around 5 to 15% of HL patients present with bone marrow (BM) involvement, as assessed by conventional biopsy.5-7 According to the Ann Arbor classification, BM involvement reflects a high tumour burden and therefore defines stage IV disease.8 Positron emission tomography (PET) using 18F-fluoro-2-deoxy-d-glucose (FDG) has demonstrated superior sensitivity compared to BM biopsy.9-11 In paediatric advanced HL, only individuals showing three or more BM lesions were assumed to be at higher risk for progression or relapse.12 Yet, there are insufficient data available to establish whether this also holds true for adult patients receiving intensive first-line treatment.

We performed a retrospective analysis based on individuals who had undergone PET at baseline and after two cycles of chemotherapy (PET-2) in the German Hodgkin Study Group HD18 trial. The aim of our study was to assess the prognostic impact of different BM involvement patterns on early metabolic response and progression-free survival (PFS) among subjects treated with PET-guided eBEACOPP. Details of the HD18 trial design and methods used for this analysis are provided as Supplementary Material.

Of the 1945 patients from our intention-to-treat population, 424 individuals with available baseline PET were eligible for the present study. Median age and sex distribution as well as PET-2 response and PFS did not differ in subjects with and without PET staging (Table S1). A total of 124 patients (29%) were BM PET-positive (BMPET+), including 56 with one or two sites of skeletal uptake (13%) and 68 showing three or more lesions (16%). BM involvement was histologically confirmed in one out of 56 individuals who had one or two sites of skeletal uptake and in 18 of the 68 patients with three or more lesions. Of the 300 BM PET-negative (BMPET−) patients (71%), 231 showed physiological (54%) and 69 diffuse BM uptake (16%). One of the individuals with diffusely increased FDG uptake in the skeleton had biopsy-confirmed BM involvement. Risk factors associated with BM involvement and the distribution of skeletal lesions are presented in Tables 1 and S2, respectively.

TABLE 1. Patient characteristics according to initial bone marrow (BM) status PET-negative BM (N = 300) PET-positive BM (N = 124) p * Age, years Median 31 34 0.086 Range 18–60 18–60 Sex Female 124 (41) 42 (34) 0.16 Male 176 (59) 82 (66) Recruiting country Germany 180 (60) 70 (56) 0.18 Switzerland 60 (20) 34 (27) Austria 27 (9) 5 (4) Netherlands 22 (7) 8 (6) Czech Republic 11 (4) 7 (6) Month of enrolment Median September 2011 March 2011 0.16 Range May 2008–July 2014 October 2008–June 2014 Hodgkin lymphoma subtype Classical 219/231 (95) 88/93 (95) 1.0 Nodular lymphocyte-predominant 12/231 (5) 5/93 (5) Ann Arbor stage II 61 (20) 8 (6) <0.0001 III 161 (54) 28 (23) IV 78 (26) 88 (71) Clinical risk factors B symptoms 160 (53) 83 (67) 0.013 Large mediastinal massa 110 (37) 21 (17) <0.0001 Extra-nodal disease 59 (20) 29 (23) 0.43 Involvement of three or more nodal areas 256 (85) 116 (94) 0.022 Elevated ESRb 157 (52) 78 (63) 0.041 IPS factors Albumin <4 g/dl 152 (51) 67 (54) 0.59 Haemoglobin <10.5 g/dl 33 (11) 22 (18) 0.079 WBC count ≥15 × 109/l 60 (20) 24 (19) 1.0 Age ≥45 years 55 (18) 26 (21) 0.59 Male sex 176 (59) 82 (66) 0.16 Lymphocyte count <0.6 × 109/l or 8% of differential 22 (7) 7 (6) 0.67 IPSmod 0–1 148 (49) 54 (44) 0.0708 2–3 140 (47) 58 (47) 4–6 12 (4) 12 (10) Note Data are presented as n or n/N (column percentage) unless specified otherwise. Abbreviations: ESR, erythrocyte sedimentation rate; IPS, International Prognostic Score; IPSmod, IPS excluding Ann Arbor stage IV; PET, positron emission tomography; WBC, white blood cell.

After 2 × eBEACOPP, 79 BMPET− (26%) and 26 BMPET+ patients (21%) had a Deauville score of 4 or higher on PET [odds ratio (OR) 0.74, 95% confidence interval (CI) 0.45–1.23, p = 0.25]. Similar results could be observed in the multivariate regression analysis, accounting for other factors potentially influencing PET positivity (Table S3). In our BMPET+ cohort, nine of the individuals with one or two sites of focal uptake (16%) were still PET-positive after 2 × eBEACOPP compared to 17 of those who had three or more BM lesions (25%, OR 1.74, 95% CI 0.71–4.28, p = 0.23). Additionally, a positive PET-2 was documented in 22 patients showing diffuse BM FDG uptake (32%) and 57 subjects without increased BM uptake (25%, OR 1.43, 95% CI 0.79–2.57, p = 0.23).

The resulting five-year PFS rates were 90.7% (95% CI 87.2–94.2) for BMPET− individuals and 91.8% (95% CI 86.6–97.0) in BMPET+ patients [univariate hazard ratio (HR) 0.82, 95% CI 0.38–1.74, p = 0.6; multivariate HR 0.66, 95% CI 0.30–1.47, p = 0.31; Figure 1]. Among BMPET+ subjects, those with one or two BM lesions had a PFS probability of 85.6% (95% CI 75.7–95.6) compared to 96.8% (95% CI 92.5–100) in patients showing three or more lesions (HR 0.22, 95% CI 0.05–1.10, p = 0.065). Within the BMPET− subgroup, five-year PFS rate was 92.1% (95% CI 88.3–95.8) for individuals who had low skeletal FDG uptake and 86.2% (95% CI 77.8–94.6) in patients showing diffusely increased uptake (HR 1.84, 95% CI 0.82–4.09, p = 0.14).

Five-year progression-free survival (PFS) rate by bone marrow (BM) status on positron emission tomography (PET)

Our study on the prognostic value of BM involvement in advanced-stage HL yielded the following findings: first, individuals with PET-positive BM lesions do not have an increased risk for PET-2 positivity when undergoing treatment with eBEACOPP, irrespective of the observed pattern. Additionally, BM involvement did not show any negative effects on five-year PFS and therefore cannot be considered as a predictor of treatment failure or disease recurrence in patients receiving intensive first-line therapy.

The high number of PET-2-negative subjects translated into an excellent five-year PFS for both BMPET− and BMPET+ individuals. Importantly, PFS was also not impaired in the subgroup of patients with three or more PET-positive BM lesions. Our finding is of particular significance, as uni- and multifocal BM involvement have previously been linked to poor outcome in adults treated with adriamycin, bleomycin, vincristine, and dacarbazine or BEACOPP-like therapy regimens.13

With PET-guided eBEACOPP, advanced-stage BMPET+ individuals face PFS rates similar to those of early-favourable patients.14 In our current analysis, BM involvement did not appear to have any negative impact on outcome, either after 2 × eBEACOPP or in the long term. Hence, this conventional risk factor seems to be overcome by effective treatment strategies aiming for cure on the first attempt. However, the findings do not provide grounds for further intensification of treatment in advanced-stage BMPET+ patients even when they have multiple skeletal lesions. On the other hand, individuals with limited or no BM involvement should not be de-escalated upfront, as they are still at risk of PET-positive residual disease after 2 × eBEACOPP. This underscores the difficulty of defining clinical risk factors for treatment failure under highly effective therapy approaches.

Our study raises the question of whether a conventional BM biopsy should still be performed routinely in HL patients with focal skeletal FDG uptake. PET imaging was shown to identify BM lesions at a higher sensitivity than biopsy of the iliac bone and has become part of diagnostic work-up in most countries.15 Current study protocols no longer recommend BM biopsy in PET-negative individuals. Based on the excellent early tumour response and five-year PFS revealed by our current analysis, biopsy-based confirmation should be restricted to subjects in whom BM involvement is leading to upstaging and subsequently more intensive treatment.

TRIAL REGISTRATION

ClinicalTrials.gov identifier – NCT00515554.

ACKNOWLEDGEMENTS

HD18 was funded by German Cancer Aid grants no. 107957 and 110617, the Swiss State Secretariat for Education, Research and Innovation as well as Roche Pharma AG grant no. ML-21683. We are grateful to all participating patients, their families, and the site investigators.