Early Positron Emission Tomography Response – Adapted Treatment in Stage I and II Hodgkin Lymphoma: Final Results of the Randomized EORTC/LYSA/FIL H10 Trial

,


INTRODUCTION
Treatment of early-stage Hodgkin lymphoma (HL) is extremely successful, with a more than 90% cure rate.The backbone of these results is a combination of chemotherapy and radiotherapy (RT) 1,2 ; however, late toxicities-mainly second malignancies 3,4 and cardiovascular events-are of major concern. 5,6tended RT fields and alkylating agents are held responsible, and avoidance of RT has been tested in several randomized trials but is associated with an increased risk of early relapse. 7onversely, a small, clinically relevant subgroup of patients experience failure with standard combined modality treatment (CMT) treatment.A sensitive tool to identify early those patients who have a high cure rate with current standards and those who should switch to other treatment is a clear unmet need.
][10] It is tempting to consider the ePET scan as the missing tool in selecting patients who can be treated less intensively, without RT, and those who need early intensified therapy; therefore, we designed a randomized trial in which patients with stage I and II HL were randomly assigned to either a standard CMT program, regardless of the result of the ePET, or to an experimental arm in which patients with a negative ePET scan were treated with chemotherapy alone and those with a positive ePET scan changed from conventional chemotherapy to an intensified schedule.This H10 Intergroup trial (European Organisation of Research and Treatment of Cancer [EORTC], Lymphoma Study Association, and Fondazione Italiana Linfomi) is the first and only trial, to our knowledge, to incorporate a randomized ePET response-adapted treatment strategy for both the ePET-negative and the ePET-positive patients with stage I and II HL.Results of the preplanned interim analysis of the ePETnegative groups of patients have been published. 11Here, we report the final analysis of all patients.

Eligibility
Previously untreated patients, age 15 to 70 years, with classic supradiaphragmatic stage I and II HL were eligible.Both favorable (F) and unfavorable (U) patients, according to EORTC criteria, 1 were included (U: at least one of the following criteria: age $ 50 years or more than three nodal areas or mediastinal-thoracic ratio $ 0.35 or no B symptoms and erythrocyte sedimentation rate [ESR] $ 50 or B symptoms and ESR $ 30; F: all others).Staging was based on conventional computed tomography scanning 12 ; a baseline PET was recommended.The study was approved by the scientific and ethical committees and all patients gave written informed consent (full protocol in the Appendix SA1, online only).

Study Design
In this multicenter, phase III trial, patients were randomly assigned upfront in a 1:1 ratio to either standard CMT or experimental ePET response-adapted treatment.A minimization technique was used for treatment allocation in the F and U groups separately, stratifying by institution, Ann Arbor stage (I v II), and availability of baseline PET.All patients received two cycles of doxorubicin, bleomycin, vinblastine, and dacarbazine (ABVD), after which an ePET was performed.
The primary objective of the study was to evaluate whether involvednode RT (INRT) could be omitted without loss of efficacy in ePET-negative patients after two cycles of ABVD.In the standard arm, ePET-negative patients received one (for F) or two (for U) additional ABVD cycles, followed by INRT.In the experimental arm, ePET-negative patients received two (for F) or four (for U) additional ABVD cycles but no radiotherapy (Fig 1). 11This primary objective was evaluated in F and U patients separately.
The secondary objective was to evaluate whether intensification with escalated doses of bleomycin, etoposide, doxorubicin, cyclophosphamide, vincristine, procarbazine and prednisone (BEACOPPesc) after two cycles of ABVD would improve outcome in ePET-positive patients.In the standard arm, ePET-positive patients received one (for F) or two (for U) additional ABVD cycles, followed by INRT, whereas in the experimental arm, ePET-positive patients (both F and U) switched to two cycles of BEACOPPesc followed by INRT (Fig 1).For this objective, ePET-positive F and U patients were pooled because of their presumed common poor prognosis.

Treatment
ABVD and BEACOPPesc were administered as described previously. 13RT was delivered according to the EORTC INRT concept: only initially involved nodes were irradiated instead of whole areas. 14,15Training sessions were organized before the start of the trial in all participating countries, and a nonmandatory prospective quality assurance program was performed in France.

PET Scanning
ePET was scheduled between days 22 and 25 of the second cycle of ABVD.Images were scored according to International Harmonization Project criteria. 16According to these criteria, mediastinal blood pool activity is recommended as reference background activity to define PET positivity of a residual mass $ 2 cm in greatest transverse diameter, regardless of its location.A smaller residual mass or a normal-sized lymph node is considered positive if its activity is above that of the surrounding background.A prospective central review of ePET was planned.For the Lymphoma Study Association, central review was performed online from the start of the trial in 2006. 17For EORTC and Fondazione Italiana Linfomi, central review started 2 years later as a result of information technology system incompatibility problems.In the absence of a timely centralized reading, local ePET reports decided further treatment in the experimental arm.

Statistical Design
Primary end point was progression-free survival (PFS), definedfrom the date of random assignement to date of progression-as experiencing relapse after previous complete remission or progression after reaching partial remission (50% decrease and resolution of B symptoms and no new lesions); progressive disease (50% increase from nadir of any previous partial remission lesions or appearance of new lesions) on computed tomography scan measurements during protocol treatment; or death from any cause, whichever occurred first.This objective was assessed via noninferiority test using a log-rank test stratified by stage (I v II) and availability of baseline PET, with a one-sided significance level of .025.

jco.org
Assuming a 5-year PFS in the ePET-negative standard arm of 95% for F group and 90% for U group, as well as allowing for a decrease of 10%, the resulting noninferiority margins for the hazard ratios (HRs) were 3.2 and 2.1 for F and U groups, respectively.As a result of the good prognosis of these patients, the absolute difference of 10% translates into noninferiority margins that are markedly higher than the noninferiority margins for HRs.Noninferiority is concluded if the upper bound of the 95% CI for the estimated HR does not exceed the noninferiority margin.A total of 26 and 63 events (progressions or deaths) for F and U groups, respectively, were required for 80% power.The original protocol estimated that a total of 248 ePET-positive patients and a total of 77 events would be obtained simultaneously.This would give 77% power to detect a 20% improvement in 5-year PFS (from 50% to 70%) using a two-sided log-rank test (a = .05).
The interim futility analysis of ePET-negative patients was previously reported and a safety amendment to close the ABVD only arms was issued in August 2010. 11At the same time, an interim efficacy analysis was performed on ePET-positive patients and the target accrual for ePETpositive patients was increased to 355 to reach the expected number of events in a reasonable timeframe.The study was closed without any interruption in recruitment in June 2011.In 2013, the independent data monitoring committee recommended that the final analysis be performed at least 3 years after the last patient was enrolled.

Statistical Analysis
Primary efficacy analyses were conducted on the intention-to-treat (ITT) principle in all patients with ePET result after the first two cycles of ABVD.Preplanned sensitivity analyses, including a per-protocol analysis, were also performed.The present noninferiority analysis on ePET-negative patients includes 900 patients who were included in the interim analysis 11 as well as additional patients who were randomly assigned between the interim analysis and the safety amendment (overruns).ePET-negative patients included after the safety amendment and all who were treated with CMT are not included in this final analysis.A Cox proportional hazards regression model was used to estimate treatment effect HRs and their 95% CIs.Survival curves were estimated on the basis of the Kaplan-Meier method and 5-year survival estimates were provided, together with their 95% CI.In ePET-positive patients, a two-sided superiority test for PFS was used with a significance level of .037 to account for the type I error spent at the interim analysis for early superiority.

Patients
From November 2006 to June 2011, 1,950 patients were enrolled (Fig 2).Clinical characteristics of 361 ePET-positive patients, 1,059 ePET-negative patients who were treated per the initial protocol, and 505 ePET-negative patients who were enrolled after the safety amendment are listed in Table 1.Twenty-five patients who were included did not receive ePET after two cycles of ABVD.Baseline recommended PET was performed in 96% of patients.Seventy-five percent of ePET scans were centrally reviewed, with 93% concordance with local assessment (Cohen's k = 0.78; 95% CI, 0.74 to 0.82).ePET positivity was reported in 18.8% of patients: 13.0% and 22.4% in F and U groups, respectively.

Outcome of ePET-Positive Patients
In the overall ePET-positive group (n = 361), after a median follow-up of 4.5 years, a total of 57 events for PFS occurred: 41-36 relapses and 5 deaths not related to HL-in the ABVD + INRT arm Patients were randomly assigned (N = 1,950) Preplanned sensitivity analyses, including a per-protocol analysis, were performed and led to similar conclusions (Appendix Table A1, online only).

Outcome of ePET-Negative Patients
In the F group (n = 465), after a median follow-up of 5.0 years, a total of 33 events for PFS occurred: two patients experienced relapse in the ABVD + INRT arm versus 30 patients who experienced relapse and one patient who died from a cause not related to HL in the ABVD only arm.ITT 5-year PFS rates were 99.0% (95% CI, 95.9 to 99.7) and 87.1% (95% CI, 82.1 to 90.8) in the ABVD + INRT and ABVD only arms, respectively, with HR, 15.8 (95% CI, 3.8 to 66.1) in favor of ABVD + INRT (Fig 4A).Noninferiority could not be demonstrated as the upper bound of the 95% CI for the estimated HR (66.07) exceeded the prespecified noninferiority margin (3.2).
In the ABVD + INRT arm, no patient experienced relapse exclusively in previously involved nodes, one in previously uninvolved nodes, and one in both, whereas in the ABVD only arm, 22 patients experienced relapse exclusively in previously involved nodes, five in previously uninvolved nodes, and three in both.In the U group (n = 594), after a median follow-up of 5.1 years, a total of 54 events for PFS occurred: 16 patients experienced relapse and six died from causes not related to HL in the ABVD + INRT arm, whereas 30 patients experienced relapse and two died from causes not related to HL in the ABVD only arm.ITT 5-year PFS rates were 92.1% (95% CI, 88.0 to 94.8) and 89.6% (95% CI, 85.5 to 92.6) in the ABVD + INRT and ABVD only arms, respectively, with HR, 1.45 (95% CI, 0.8 to 2.5) in favor of ABVD + INRT (Fig 4B).Noninferiority could not be demonstrated as the upper bound of the 95% CI for the estimated HR (2.50) exceeded the prespecified noninferiority margin (2.10).
In the ABVD + INRT arm, five patients experienced relapse exclusively in previously involved nodes, four in previously uninvolved nodes, and six in both, with one unknown, whereas in the ABVD only arm, 20 patients experienced relapse exclusively in previously involved nodes, four in previously uninvolved nodes, and six in both.
Causes of death are listed in Table 2.The 5-year OS rates were 96.7% versus 98.3% for ABVD + INRT and ABVD only arms, respectively (Appendix Fig A1B).Preplanned sensitivity analyses, including a per-protocol analysis, were performed and led to similar conclusions (Appendix Table A1).
The outcomes of the 505 ePET-negative patients who were included after the safety amendment and all treated with CMT did not differ from those of the standard CMT in the randomized setting (Table 2).

Treatment Compliance and Toxicity
No unexpected toxicities were observed in ePET-negative patients in either arm.

DISCUSSION
The H10 trial was designed to evaluate an ePET response-based adaptation of treatment in patients with stage I and II HL.If ePET-positive patients could benefit from early intensification of chemotherapy, failure rates could decrease.If ePET-negative patients could be spared RT, long-term toxicity could be reduced.
After two cycles of ABVD, 18.8% of patients had a positive ePET.A significant improvement (13.2%) of 5-year PFS was reached in the experimental BEACOPPesc + INRT arm compared with continuation with ABVD + INRT.Even a benefit in OS of 6.7%, with a trend toward statistical significance, was observed.This finding is of immediate clinical relevance because on the basis of an ePET result, , 20% of patients needed intensified chemotherapy to obtain superior PFS rates after starting with ABVD.In the German Hodgkin Study Group HD 14 trial, patients with intermediate stage I and II HL were randomly assigned to either two cycles of BEACOPPesc followed by two cycles of ABVD versus four cycles of ABVD in both arms followed by involved-field RT (IFRT). 18No ePET was included in this design.This resulted in a modestly but significantly better PFS for BEACOPPesc + ABVD, but at the cost of exposing all patients from the start to BEACOPPesc.The H10 results demonstrate that ePET response-adapted intensification is as effective as the HD14 approach, but might spare 77.6% of the U patients from treatment with BEACOPPesc.The relative dose intensity of the restricted number of two cycles of BEACOPPesc was excellent and hematologic toxicities were manageable.We do not yet have results on fertility in our trial, but preliminary analysis of the HD 14 trial revealed no significant differences in female fertility potential after two cycles BEACOPPesc and two cycles of ABVD compared with four cycles of ABVD. 19The gap in time between random assignment and experimental intervention could be associated with a risk of bias in the current study; however, clinical characteristics and prognostic factors seem to be adequately balanced between both arms.In view of the significant reduction in the risk of relapse, manageable toxicity, acceptable risk of female infertility, and avoidance of second-line high-dose treatment with autologous stem cell rescue for patients who experience relapse, PET-adapted intensification in ePET-positive patients is a realistic treatment option.
ePET-negative patients had excellent outcomes, but the 5-year risk difference in PFS was 11.9% and 2.5% in favor of CMT in the F and U groups, respectively; however, for both the F and U groups, noninferiority of ABVD only could not be declared as the upper bound of the 95% CI of the estimated HRs exceeded  jco.org the preset noninferiority margins, which was in line with the conclusions of the interim analysis. 11Our decision to consider a 210% difference in 5-year PFS rates can be questioned.It reflects our attempt to balance immediate high cure rates and the risks of serious long-term toxicities. 4,5In the RAPID trial, which was designed to evaluate whether IFRT could be omitted when the PET scan after three cycles of ABVD was negative, comparable assumptions (27% margin at 3 years) were adopted to balance cure rates and toxicities. 20e observed some remarkable differences between the F and the U groups.In the F group, INRT still improved disease control even in ePET-negative patients, but the overall excellent survival in both arms supports the consideration of INRT omission for selected patients in the context of individualized therapy.In the U group, 5-year PFS was only modestly impaired after omission of INRT, mainly because of several events after .36 months in the standard arm.This finding challenges the need for RT more explicitly in the U group.The less prominent role of INRT in patients in the U group compared with those in the F group is not evident, but perhaps the number of additional chemotherapy cycles after ePET negativity in the ABVD only arms might have played a role-two additional cycles of ABVD in the F group versus four in the U group.This assumption is supported by the analysis of relapse patterns.For patients who were treated in the experimental arms with ABVD only, relapses occurred predominantly in previously involved nodes, mainly early after end of treatment (# 2 years).In the standard CMT arms, INRT seemed to be successful in preventing relapses in initially involved nodes as only 15.6% of patients who experienced relapse developed disease in these sites.However, in the U group, which had more widespread disease than the F group, late relapses (. 3 years after end of treatment) occurred more frequently after CMT than after ABVD only, which possibly reflects a suboptimal amount of chemotherapy in the CMT arm.In the RAPID trial, including mostly favorable patients and excluding mediastinal bulk, no additional chemotherapy was administered to PET-negative patients after three cycles of ABVD, and a comparison was made between no further treatment and IFRT. 20Here, the difference in timing of relapse between patients who received CMT and those who received chemotherapy only was not observed, which possibly reflects the influence of additional cycles of chemotherapy in the H10 trial.Long-term toxicities could not yet be incorporated in the interpretation of the outcome.This hampers the actual clinical judgment of noninferiority of a new treatment approach but is inevitable in this group of long-term survivors.
Prognostic factors, such as ESR, B symptoms, age, and bulky mediastinum defined on a chest x-ray and used in the current study, are clearly challenged in the current era of ePET-directed therapy.The observation in the current study that 377 (72.9%) of 517 of the bulky mediastinum (mediastinal-thoracic ratio $ 0.35) are PET negative after only two cycles of ABVD suggests that ePET may overcome some classic prognostic factors in localized HL as it overcomes the International Prognostic Score for advanced HL. 9 In stage I and II HL, the outcome of patients is excellent with CMT.To improve the balance between efficacy and toxicity, ePET helps in the treatment decision for individual patients.The H10 shows that when ePET is positive after two cycles of ABVD, intensification with two cycles of BEACOPPesc + INRT should be considered as the best treatment option.In ePET-negative patients, the overall outcome is excellent: either after CMT or after chemotherapy only.But statistically-on the basis of our preset assumptions-noninferiority of leaving INRT out of the treatment could not be demonstrated; a negative ePETseems not to be the ideal tool to identify patients who do not need RT.In the F group, CMT results in a better immediate disease control; however, in the U group, benefit of CMT seems to be less clinically relevant, and treatment with chemotherapy only is defensible in individual patients.

Fig 4 .
Fig 4. Progression-free survival of 1,059 early positron emission tomography-negative patients who were treated per the initial protocol.Shown are the rates of progression-free survival of the (A) favorable (F) groups of patients randomly assigned to doxorubicin, bleomycin, vinblastine, and dacarbazine (ABVD) + involved-node radiotherapy (INRT; n = 227) or ABVD only (n = 238) and of the (B) unfavorable (U) groups randomly assigned to ABVD + INRT (n = 292) or ABVD only (n = 302).HR, hazard ratio, O observed; n, number of patients.

Table 1 .
Patient Characteristics : all others †All early PET-negative patients who were included after the safety amendment received treatment with ABVD + INRT, as the independent data monitoring committee recommended to close the ABVD only arm.‡Bulkymediastinum is defined as a mediastinal-thoracic ratio $ 0.35.§As the median follow-up is shorter in these groups that were included after the safety amendment, PFS and OS are given at 3 years.jco.organd16-13relapses and 3 deaths not related to HL-in the BEACOPPesc + INRT arm.ITT 5-year PFS rates were 77.4% (95% CI, 70.4 to 82.9) and 90.6%(95% CI, 84.7 to 94.3)in the ABVD + INRT and BEACOPPesc + INRT arms, respectively, with HR, 0.42 (95%CI, 0.23 to 0.74; P = .002)in favor of BEACOPPesc + INRT (Fig 3A).

Table 2
lists the causes of death.The 5-year OS rates were 100.0%versus 99.6% for ABVD + INRT and ABVD only arms, respectively (Appendix Fig A1A, online only).