Management of early-stage Hodgkin lymphoma: is there still a role for radiation?

More than a century has passed since it was first shown that Hodgkin lymphoma could be effectively treated by irradiation¹  and it has been known for many decades that this approach can be curative, especially for patients with localized disease.^2,3  Indeed, the efficacy of radiation as a “single agent” is probably greater than that of any individual cytotoxic drug. Nonetheless, it is also clear from many trials that the results of treatment for early-stage disease are improved by the combination of chemotherapy with radiation. The principle of combination treatment has led to the greatest advances in treating classical Hodgkin lymphoma, but the incorporation of radiotherapy into such combinations is falling from favor in the minds of many because of concerns about its toxicity. To what extent is this retreat from radiation justified by the evidence from clinical trials? This review examines the results of previous trials and the emerging data from several recent studies that have sought to use functional imaging to guide a more selective approach using measurements of the response to chemotherapy to determine whether to use radiotherapy.

The initial use of radiotherapy for Hodgkin lymphoma was based upon extensive fields of treatment covering the known sites of involvement, as well as lymphatic groups in the same region of the body, on the premise that Hodgkin lymphoma progresses in an anatomically coherent fashion from node to node and that seemingly uninvolved nodes may harbor subclinical deposits that can be eliminated by irradiation.⁴  For the most common presentation, with disease in the thorax or neck, this led to the use of subtotal nodal irradiation (STNI) including the cervical, axillary, mediastinal, hilar, and paraaortic nodes and the spleen treated in sequence.⁵  The overall 10-year progression-free survival (PFS) with this approach was ∼ 80% for stage I and II disease but, with prolonged follow-up, the late toxicity of such an approach has become apparent. The dominant problems that emerge are cardiovascular disease and secondary malignancies.

Cardiovascular disease consists mainly of the accelerated development of atherosclerosis in irradiated arteries and valvular fibrosis when the heart is included in the radiation field.⁶  A dose-dependent increase is seen in the risk of congestive cardiac failure, pericardial disease, and valve abnormalities in those exposed to 15 Gy or more.⁷  There is also an apparent increase in the risk of myocardial ischaemia at lower radiation doses; a recent analysis suggested an overall hazard ratio (HR) of more than 12 for those treated with mediastinal radiotherapy in childhood.⁷  A 4-fold increased incidence of ischemic stroke has been recorded after treatment with mantle radiotherapy in childhood.⁸ 

The evidence of second malignancies related to radiation exposure continues to accumulate. The clearest risks are for epithelial cancers within or adjacent to the radiotherapy field, which show a latency of ∼ 3 years but increase progressively over time.^9,10  There is also some suggestion of an increased risk of myelodysplasia and acute myeloid leukemia with extensive irradiation, although the analyses of this are complicated by the extensive use of alkylating agents at the same time, which carry a significant risk in their own right.¹¹  The risks of secondary solid tumors are clearest for breast cancer in women given extensive radiotherapy to the thorax including the axillae. The relative risks are highest for those irradiated around the age of puberty and decrease progressively with older age at treatment, although the rising background incidence of breast cancer in older groups results in a less steep decline in absolute risk with age.¹²  The excess risk appears to approach zero for those irradiated at age 40 or older, but the results between 30 and 40 years of age are less clear, with some series showing a significantly increased risk in those irradiated between 30 and 35 years, which includes the peak age of incidence of Hodgkin lymphoma among adults. The time taken to develop breast cancer seems to be slightly shorter for those irradiated younger, the relative risk peaking at 10 to 14 years for those irradiated before the age of 20 and 15 to 19 years for those treated later, although in both groups the absolute excess risk is greatest after ∼ 30 years. In a large study from the United Kingdom, the total cumulative risk reached nearly 50% for those treated under the age of 20 with at least 40 Gy, most of which occurred beyond 20 years of follow-up.¹²  Similar trends are found for lung cancer and mesothelioma in those who receive mantle radiotherapy, especially among smokers,¹³  and, among children, the incidence of subclinical hypothyroidism and thyroid cancers is also substantial after cervical irradiation.¹⁴ 

These studies have shown a consistent relationship between size of field and/or dose of radiation for the risks of solid tumors, so we can expect that the move to lower dose and more limited fields will reduce the risks for patients treated with modern radiotherapy techniques.¹⁵  Moving from extended-field (EFRT) to involved-field radiotherapy (IFRT) has been shown to result in similar overall survival (OS) despite a slight excess of recurrent disease when radiotherapy was used alone for early-stage Hodgkin lymphoma.^16,17  A series of trials have demonstrated that using chemotherapy in a combined modality approach will produce superior PFS with smaller radiotherapy fields compared with EFRT alone (Table 1).^18-21  On this basis, there is no longer any role for EFRT as the sole treatment for early Hodgkin lymphoma and the impetus now is toward further restriction of the fields and lowering doses in combined modality treatments.

The increasing sophistication of cross-sectional imaging and conformal radiotherapy planning has allowed a progressive reduction in field size, moving to treatment of only the nodes involved by disease rather than the whole node group. The definition of the precise field for this involved-node irradiation remains a matter of some debate, in particular the margins required to allow for movement during treatment and whether functional imaging using 2-(¹⁸F)fluoro-2-deoxy-D-glucose positron emission tomography (FDG-PET) may allow greater precision.²²  However, from preliminary data, this approach appears to be equally effective on the basis of cohort studies using historical controls,²³  and a prospective randomized trial to compare involved-node irradiation and IFRT is in progress in the German Hodgkin Study Group (GHSG; http://www.controlled-trials.com/mrct/trial/1717367/HD17). In the future, different techniques such as the use of 3D proton beam therapy may allow further restriction of the exposure in normal tissues such as the heart, lungs, and breasts.

In addition to reducing the size of radiotherapy fields, several studies have been performed to identify the lowest effective dose of radiation. Historically, doses of 36 Gy and higher were used for the consolidation of remission in early-stage disease,^18-20  but modern practice is to apply 30 Gy or less in some circumstances. Two key studies in this respect have been reported by the GHSG, in early favorable disease (HD10²⁴ ) and unfavorable disease (HD11²⁵ ), respectively. Early favorable disease is defined by the absence of adverse risk factors, specifically erythrocyte sedimentation rate > 50 or > 30 with B symptoms, extranodal disease, more than 2 sites of involvement, or mediastinal bulk disease. In the HD10 study, a 2×2 factorial design used randomization between 2 and 4 cycles of ABVD (doxorubicin, bleomycin, vinblastine, dacarbazine), followed by randomization between 20 and 30 Gy of consolidation IFRT. The results in all the groups were excellent and superimposable, with a projected 8-year freedom from treatment failure (FFTF) of 86% and an OS of 95% in the group who received only 2 ABVD and 20 Gy. In HD11, a similar design was used, but randomizing these patients with adverse features between 4 cycles of ABVD and 4 cycles of BEACOPP (bleomycin, etoposide, doxorubicin, vincristine, procarbazine, prednisolone, followed by 20 or 30 Gy of IFRT. In this case, although the OS was equally good in all groups (95% at 5 years), the patients treated with only ABVD and 20 Gy showed inferior FFTF (81% vs 87% in the other groups). This suggests an interaction between the intensity of chemotherapy and the minimum effective dose of radiotherapy: a lesser dose of radiation may be compensated by more intensive chemotherapy or vice versa, but when both are reduced, the control of the lymphoma is also lessened. This is an important point, because when the question becomes the complete omission of radiotherapy, the intensity of the chemotherapy is likely to become an even more dominant consideration, at least in the initial control of the disease.

The other important point from this and many other studies in Hodgkin lymphoma is the disconnect between FFTF and OS: the very good results of salvage treatment permit cures in the great majority of patients whose initial treatment fails. In calculating the optimum strategy, there is a balance to be struck between subjecting all patients to maximum-intensity treatment in the first line and recognizing that the use of less intense treatment with a slightly higher initial failure rate may be preferable if it avoids serious late toxicity and mortality. The relative risks of death from Hodgkin lymphoma or from the complications of treatment are a key consideration in selecting a regimen, with much current research aimed at determining where the balance lies and how both might be minimized simultaneously.

The natural extension of this thinking, after the success of combined modality therapy, is to examine the use of combination chemotherapy alone in patients perceived to be at low risk of treatment failure but high risk of radiation-induced toxicity (Table 2). A series from 1 center suggested that 6 cycles of ABVD was effective for early-stage Hodgkin lymphoma, with only 6 recurrences among 71 subjects with early favorable disease under the age of 45 and no deaths at a median follow-up of 5 years,²⁶  although it is hard to interpret such data in the absence of a control group. A small randomized study in nonbulky stage I, II, and IIIA disease examined 6 cycles of ABVD with or without consolidation IFRT, with no significant difference in outcome among 152 patients entered, although there was a trend to less favorable results with chemotherapy alone, which did not reach statistical significance.²⁷ 

There are particular reasons to avoid radiotherapy if possible among children and young adults with Hodgkin lymphoma. These include not only the higher cumulative risks of secondary malignancy and cardiovascular disease, but also the effects upon skeletal growth and maturation. For this reason, 2 pediatric oncology groups have examined chemotherapy-only approaches in randomized trials. In the North American Children's Oncology Group study CCG 5942, patients reaching complete remission at the end of chemotherapy were randomized to low-dose (21 Gy) IFRT or no further therapy.²⁸  The trial stopped accrual early when an interim analysis revealed a significant difference in the progression event rates in the 2 arms. A total of 498 patients received 4 cycles of COPP-ABV chemotherapy before randomization. With a median 7.7 years of follow-up, there was a significant difference in event-free survival favoring the radiotherapy group (93% vs 83%, P = .004). Interestingly, ∼ 90% of recurrences occurred in sites of original disease, which would have been irradiated in the other trial arm. There was, however, no difference in OS, with 10-year estimated survival rates of 97% and 96%, respectively, across all risk groups. When analyzing patients according to risk groups defined by stage and the presence of risk factors, the only subgroup to show a significant difference in event-free survival was the favorable one, in which patients with early-stage disease received less chemotherapy than the groups with risk factors such as bulky disease or advanced stage. This difference might be due to a preferential effect of radiotherapy when the disease is localized or may be further evidence supporting an interaction between the value of radiotherapy and the extent of prior chemotherapy.

Another study in younger patients was conducted by the German Pediatric Oncology and Hematology group (HD95). In that study, those with complete remission at the completion of chemotherapy were managed expectantly, whereas those with residual disease went on to receive IFRT on a dose scale from 20 Gy for those in good partial remission up to 35 Gy for masses more than 50 mL.¹⁴  Patients with early-stage favorable disease received only 2 cycles of the vincristine, procarbazine, prednisone, and doxorubicin (OPPA) and vincristine, etoposide, prednisone, and doxorubicin (OEPA) regimen, and the results with a median 10 years of follow-up were excellent: PFS of 97% and OS of 98.5% among 66 patients not treated with radiotherapy. In that study, it was the intermediate-risk group with stage IIB or IIIA disease or extranodal involvement that had less favorable results after the omission of IFRT, with only a 68% 10-year PFS. Only one secondary malignancy was seen among the 165 patients who did not receive radiotherapy, whereas 17 occurred among 746 patients who had radiotherapy (of these, 14 arose within the radiation fields) and 30% of those who received radiotherapy to the neck developed hypothyroidism.

The largest study to make a direct comparison of chemotherapy alone with a combined modality approach was the intergroup study HD.6 of 399 patients with nonbulky stage I and II disease.²⁹  This compared treatment with ABVD alone to either 35 Gy STNI alone in favorable early disease, or 2 cycles of ABVD followed by STNI in unfavorable cases. The early report of this study, with a median follow-up of 4.2 years, showed that the chemotherapy-alone groups had an inferior freedom from progression of 87% versus 93% (P = .006), with the unfavorable group particularly disadvantaged by the omission of radiotherapy (88% vs 95%, respectively; P = .004). At that time, the OS was very good and not significantly different between the arms (94% vs 96%, respectively), but a subsequent reanalysis with follow-up of 11.3 years showed a different picture, with inferior survival among the patients who had received radiotherapy (87% vs 94%, respectively; P = .04).³⁰  The risk of death from Hodgkin lymphoma was not different between the arms, but the risk of death from other causes was more than 3-fold higher among those irradiated, with much of the excess attributable to second cancers. The present-day relevance of this study is limited by its use of EFRT, an approach that is no longer considered appropriate because it increases the irradiated volume 3- to 5-fold and includes the mediastinum, with the risk of cardiac disease; the axillae, with the risk of breast cancer; and the spleen, with the risk of opportunistic infections. Perhaps the most interesting finding is that the control of Hodgkin lymphoma and OS can be influenced in opposite directions such that the risk of progression was approximately double in the chemotherapy alone arm, whereas the risk of death was halved. This highlights that freedom from progression is a poor surrogate end point or proxy measure for the ultimate outcome of treatment.

To make a comparison between the best combined modality therapy and the proposed standard chemotherapy-only regimen, an individual patient meta-analysis was undertaken to compare the results in the GHSG H10 and H11 trials with the intergroup HD.6 study.³¹  In this analysis of 406 patients who fulfilled the eligibility criteria for both the intergroup study and either of the GHSG studies, combined modality therapy in the HD10 and HD11 studies appeared to deliver better time to progression than that seen using ABVD alone in comparable patients in the HD.6 study (HR = 0.44; 95% confidence interval [CI], 0.24-0.78). For the PFS, however, the difference did not reach significance (HR = 0.71; 95% CI, 0.42-1.18) and the OS was superimposable. The difference between the treatments was particularly evident among patients who did not reach complete remission with initial chemotherapy (PFS; HR = 0.35; 95% CI, 0.16-0.79), although there was still no OS difference in this group.

In view of the information described, it would be helpful to have the means to differentiate between patients likely to have been cured by chemotherapy alone and those for whom the treatment has been less effective and the risks of irradiation may be acceptable to improve the chances of cure. There is considerable interest in the possible role of FDG-PET scanning, which can be used in combination with CT scans to determine accurately the level of metabolic activity in any particular tissue mass. This gives much greater sensitivity to the presence of viable Hodgkin lymphoma in small lymph nodes and has been extensively evaluated as a means to determine the degree of response early during a course of treatment. A 5-point scale has been tested across multiple centers and shown to be reproducible in the evaluation of FDG-PET uptake³²  (Table 3). In early Hodgkin lymphoma, 3 prospective randomized studies have been performed (Figure 1), of which 2 have so far presented early results.^33,34  The designs of these studies are complementary, with different amounts of treatment planned and different points of randomization, but all intend to address the hypothesis that radiotherapy may be reserved for those patients found to have metabolically active sites of lymphoma on an interim FDG-PET scan and avoided for those in whom a good response has already been achieved. Although the follow-up is short, some differences have already emerged from the data reported.

The European Organisation for Research and Treatment of Cancer (EORTC) H10 study recruited 1137 early-stage patients across both risk groups and after 2 cycles of ABVD patients were reported as PET negative in 85% of favorable and 74% of unfavorable cases in the control arm and in 88% and 78% of cases, respectively, in the experimental arm where radiotherapy would be omitted.³⁴  With a median follow-up of just over 1 year, for the PET-negative groups, the experimental arm showed significantly inferior PFS in the favorable group (94.9% vs 100%; HR = 9.36; P = .017) and 94.7% versus 97.3% in the unfavorable group (HR = 2.42; P = .026). This suggested that the PET-negative patients who did not receive IFRT could not match the results in the PET-negative controls, all of whom had consolidation radiotherapy, and accrual was halted. There was only 1 death, from toxicity rather than lymphoma, at the time of the analysis, so data on OS will await much longer follow-up.

The UK National Cancer Research Institute RAPID study included 602 patients with nonbulky stage IA and IIA disease, of whom 2/3 had favorable disease by GHSG criteria.³³  After 3 cycles of ABVD, 75% patients had a score of 1 or 2 on the 5-point scale and were randomized to no further therapy or to 30 Gy of IFRT. With a median 4 years of follow-up, the PFS at 3 years was not significantly different between the allocated groups (94.5% vs 90.8%; P = .23). However, there were 26 patients who did not receive radiotherapy as allocated, mainly from patient or physician choice and, on analyzing the results by treatment actually received, there was a significant difference in 3-year PFS in favor of the irradiated patients (97% vs 90.7%; HR = 2.39; P = .003). There were 7 deaths in the group randomized to IFRT, but in 5 of these, radiotherapy was not given. One patient in the nonirradiated group died and the 3-year OS was not significantly different between the randomized groups.

There is a risk of overinterpretation based upon very small numbers of events in these 2 trials, but there are interesting differences between them, in particular the results of the FDG-PET scanning, with a higher proportion of patients judged PET negative in H10 than in RAPID despite the exclusion of bulky disease and one more cycle of ABVD given before the scans in the UK study. The RAPID trial used real-time prospective central review with a strict quality control system for the scan acquisition, whereas the EORTC study relied upon local interpretation of scans for immediate decisions on therapy in the experimental arm. It is difficult to know why the PET-negative rate in the EORTC study was higher in the patients on the experimental arm in both the favorable and unfavorable groups, but this may have some influence on the outcomes if borderline positive patients were being allocated to the PET-negative group. A review of FDG-PET scans performed retrospectively moved 6 patients from negative to positive, but this did not influence the conclusions in the analysis.

Irrespective of the details in these 2 studies, the HRs for PFS all point in the same direction, with a small excess of recurrences in the nonirradiated groups who were PET negative after initial ABVD. It is clear that the sensitivity of FDG-PET scanning in this setting is some way short of 100%. Despite this, the overall results in both studies are encouraging, with more than 90% of patients free from disease at 3 years and a narrow margin of inferiority that many patients and clinicians may be prepared to accept to avoid the potential long-term sequelae of radiotherapy. Much longer follow-up is clearly essential to make a true determination of the relative merits of these strategies, in particular whether there is any difference in OS with time.

The results of treatment for early Hodgkin lymphoma have improved massively over the last few decades, and we have now arrived at a point where the problems of life after cure have become almost more pressing than the cure itself. Diligent clinical investigation and the careful application of new techniques in radiation, imaging, and chemotherapy have progressively improved the balance of risk for patients to the point where the alternative approaches to treatment become a matter of personal choice. Some would opt for combined modality therapy as offering the highest chance of cure at the first attempt, while others would prefer chemotherapy alone and risk the small chance of requiring salvage therapy in the knowledge that the chances of cure at the second attempt remain very high. For the future, further refinements in our understanding of biological heterogeneity and its influence on the outcome of treatment may make these choices better informed, as may improvements in functional imaging with newer tracers. The optimization of initial therapy to drive up the complete response rate is an attractive idea, particularly in view of the emerging data with the immunotoxin brentuximab vedotin, which may provide a means to bring the PET-negative numbers higher still.³⁵  For the present, patients can be informed that the overall outlook is excellent, that there are different options for treatment that carry small competing risks, and that the best thing to do is still to take part in a clinical trial because this is how we will continue to make progress.

Conflict-of-interest disclosure: The author has received honoraria from Millennium Takeda. Off-label drug use: None disclosed.

Peter W. M. Johnson, Cancer Research UK Centre, University of Southampton, Southampton, UK, Somers Cancer Research Building, Mail Point 824, Southampton General Hospital, Southampton SO16 6YD, UK; Phone: +44 2380 796186; Fax: +44 2380 795152; e-mail: johnsonp@soton.ac.uk.