Response‐adapted therapy in Hodgkin lymphoma

The outlook for patients with classical Hodgkin lymphoma (HL) has improved dramatically over the last 60 years, and it is now considered one of the most curable forms of cancer through the use of highly effective chemotherapy, with radiotherapy in a proportion of cases. The improvements in survival rates in this relatively young patient group have exposed the problem of long-term treatment-related toxicity and prompted us to look beyond a cure-at-all-costs approach. There is a drive towards individually tailored treatment, designing therapy based upon individual prognostic variables and biomarkers, with the aim to achieve the best possible outcome whilst minimising the associated toxicity. Much attention has focused on the early assessment of response to treatment, which offers a case-specific, adaptive approach, enabling identification of patients whose treatment can safely be attenuated, and the minority with more resistant disease who could benefit from escalation to more effective, but potentially more toxic, therapy. It is only in the last 40 years, with the dramatic improvement in Hodgkin's survivorship, that the full extent of long-term toxicity in this particular group has emerged (Table 1). Studies of mortality following HL treatment show that deaths from other causes outweigh those directly related to HL1 and second malignancies are the leading cause of death among HL survivors.2, 3 The increased risk of breast cancer in patients previously treated with extended field radiotherapy has been consistently reported. One cohort study reported a 5.6-fold increased risk of invasive breast cancer in women treated with radiotherapy for HL compared with the general population. The risk is known to be sustained for at least 30 years after completion of treatment and is higher for women receiving radiotherapy before the age of 40, particularly where higher doses are used and larger volumes of breast tissue are irradiated.3 There is also evidence that radiotherapy and alkylating chemotherapy (AC) are linked to higher rates of lung and gastrointestinal malignancy in HL survivors. The risk of lung cancer is directly related to radiation dosage and the number of cycles of AC received, and prognosis for these patients is very poor, with median survival of less than 1 year following diagnosis. Acute myeloid leukaemia, peaking within 5 years of exposure to AC, is another well-recognised complication of treatment and also carries a poor prognosis. However, the relative risk is considerably lower with ABVD (doxorubicin, bleomycin, vinblastine, and dacarbazine) than with earlier regimes such as MOPP (mechlorethamine, vincristine, procarbazine, and prednisolone) or the more intensive escalated BEACOPP (bleomycin, etoposide, doxorubicin, cyclophosphamide, vincristine, procarbazine, and prednisolone), which entail higher cumulative AC doses.3 Alkylating agent-related infertility is known to occur with more intensive regimes such as eBEACOPP. The German Hodgkin Study Group reported severe menopausal symptoms in one-third of women, and oligospermia in 89% of men, following treatment with 6 cycles of eBEACOPP.4 There is also increased morbidity and mortality from ischaemic heart disease, valvular heart disease, conduction abnormalities, and anthracycline-related cardiomyopathy in survivors of HL treated with radiotherapy.5 The risk of bleomycin pulmonary toxicity is well documented and is associated with a significant decrease in 5-year overall survival (OS) from 90% in unaffected patients to 63% in those with confirmed toxicity.6 As these risks have emerged, clinicians are confronted with the question of how to reduce toxicity without losing disease control. Some measures have already been tested, including reducing radiation volumes and even omitting consolidation radiotherapy altogether following complete responses determined by standard cross-sectional imaging.7 A meta-analysis assessing breast cancer risk in HL survivors demonstrated an odds ratio of 3.25 for extended field versus involved field radiotherapy, without evidence of any increase in recurrence rates with field reduction.2 Given this precedent, the next logical question was whether the same could be done with systemic treatment. Prospective trials comparing ABVD with the more intensive eBEACOPP have found that in those receiving ABVD alone, fertility was largely preserved and the subsequent risks of secondary malignancy and leukaemia were negligible. However, there is also evidence of reduced efficacy with less intensive treatment. ABVD is associated with a 10% to 15% lower probability of failure-free survival than eBEACOPP,8 but it is important to highlight that this does not appear to lead to higher rates of mortality, with individual studies showing no significant difference in OS, due in part to effective salvage with second-line chemotherapy. Optimising the outcome for each patient with HL requires the selection of therapy to maximise the probability of durable remission, whilst keeping the risk of long-term toxicity to a minimum. With cure rates above 80% using current treatment, it is clear that treating all patients with intense, toxic therapy from the outset will benefit only a minority with more chemo-resistant disease. This calls for an approach that enables identification of those requiring more intensive therapy, and those in whom treatment can be safely de-escalated, confining the risk of toxicity to those likely to benefit the most. The historical approach to such stratification has been the use of the prognostic features at presentation. For anatomically early stage disease (Ann Arbor I and II), the groups have been divided into favourable and unfavourable according the presence of bulk, systemic symptoms, and the erythrocyte sedimentation rate. In advanced disease, the International Prognostic Score (IPS) is the accepted measure of prognosis at baseline, consisting of 7 independent risk factors based on analysis of 5000 patients under 65 years old treated before 1990. It therefore does not take into account subsequent improvements in survival, particularly in the poorer risk groups. As a result, a reanalysis of the IPS in 2012 has shown that the difference in 5-year freedom from progression between the low-risk (IPS 0-3) and high-risk (IPS≥4) groups has reduced from more than 30% to 15%.9 The IPS is therefore unsatisfactory for identifying the group of patients at greatest risk of treatment failure, who would benefit most from early intensification. There is some evidence that more-sophisticated biological stratification by gene expression profiling might help identify the worst prognostic groups. One model using 23 genes to predict OS demonstrated a 29% difference in 5-year OS between high- and low-risk groups (63 vs 92%), although these findings have not been replicated in other similar studies. Other studies have examined circulating biomarkers with the potential to predict disease response. The thymus and activation-related chemokine is highly expressed by Reed-Sternberg cells, and elevated levels were shown to correlate with tumour burden as well as other prognostic variables such as Ann Arbor stage. Studies of other factors such as IL6, ILR2, and soluble CD30 have also demonstrated an association between elevated levels and adverse outcomes, but larger prospective trials are required to validate these as predictive markers. In summary, the means to discriminate better from worse prognostic groups based on characteristics at presentation are unsatisfactory, and it is for this reason that a more dynamic measurement of prognosis has been sought. Response to treatment is clearly important in predicting long-term outcome. Patients who fail to reach complete remission with first line treatment have a much worse prognosis, and much attention has focused on the use of imaging to identify these at an early stage. Computed tomography (CT) is certainly useful in assessing morphological response, but an interim reduction in tumour size has not been shown to predict accurately long-term prognosis. This may relate to the fact that malignant cells represent only a small proportion of total tumour volume, so tumour shrinkage can lag behind response at a cellular level and may not occur until the latter stages of treatment, or beyond. Functional imaging with positron emission tomography (PET) and PET-computed tomography (PET-CT) with 2-(18F) fluoro-2-deoxy-D-glucose (FDG) has been extensively studied and, in early retrospective studies, was shown to be superior to CT and IPS in predicting durable remissions. Interim PET-CT (iPET) after 2 cycles of ABVD was associated with a high predictive value, with a 3-year failure-free survival of 95% in those with a negative iPET and 18% in the iPET positive group.10 In this setting, IPS was shown to have no significant independent prognostic value, with iPET positive patients categorised as low risk by IPS at equally high risk of relapse as those in high IPS groups. Positron emission tomography imaging relies on the increased uptake of labeled FDG by actively metabolising tissues, including those affected by active HL. One confounding factor is that numerous other tissues such as myocardium, liver, kidney, brain, brown fat, inflamed lung tissue, and regenerating bone marrow are also “FGD avid.” This has resulted in efforts to establish a standardised method of interpretation, to reduce the variability of qualitative visual assessment. The 5-point Deauville scale was developed in 2009 and grades areas of abnormally high FDG uptake by comparison with normal physiological uptake in different tissues (Table 2). The scale allows the cutoff point to be adjusted based on trial design, but as a general rule, a score of 1 to 3 is reported as negative, and 4 to 5 as positive. In an international validation study involving 6 independent reviewer, there was complete concordance in differentiating a positive from a negative scan in 82% of cases.11 The high predictive value of iPET in tandem with the validated Deauville scale12 led to its inclusion in international guidelines as the standard for response assessment in HL.13 With iPET appearing to offer reliable prognostic stratification early on during treatment, a number of prospective trials have been undertaken to assess intensification and de-escalation of therapy based on iPET results, both in early stage and advanced disease. In early stage disease, studies have focused on the omission of radiotherapy following a negative iPET. In the UK NCRI RAPID trial, patients with nonbulky stage IA or IIA HL with a Deauville score of 1 to 2 on iPET after 3 cycles of ABVD (n = 420, 74.6%) were randomised to either receive Involved field radiotherapy (IFRT) or no further treatment.14 Similarly, the European EORTC H10 study compared standard therapy with PET-guided treatment, in which patients with a negative iPET after 2 cycles of ABVD did not go on to receive Involved node radiotherapy (INRT).15 Their findings were similar, each showing a small increase in the risk of recurrence following the omission of radiotherapy, but without an associated worsening of OS. RAPID reported a 3-year PFS of 94.6% in the radiotherapy group and 90.8% for those not treated. In the H10 trial, PFS at 1 year for “unfavourable risk” patients was 97.3% in the “standard therapy” group and 94.7% in the iPET-directed cohort. After 5 years, the PFS values for these groups were 92% and 90%, indicating that IFRT delayed rather than preventing recurrence in a proportion of cases, with a significant number of relapses occurring outside the radiotherapy field. A modest improvement in PFS is offset by exposing all patients to radiation, most of whom gain no benefit and are put at risk of significant toxicity. From the published trials, the number needed to treat with radiotherapy to prevent 1 recurrence is between 25 and 30. Overall, these findings show that patients with early stage disease have an excellent prognosis with or without consolidation radiotherapy, but longer term follow-up is needed to determine the extent to which second malignancies and cardiovascular disease are avoided by this approach. EORTC H10 is the only prospective randomised study to have reported on escalation of therapy. Of the 19% of patients in the unfavourable group who were iPET-positive following 2 cycles of ABVD, 5-year PFS for those randomised to continue ABVD followed by INRT was 77%, compared with 91% in those who went on to receive eBEACOPP and INRT (hazard ratio 0.42, P = 0.002). Remissions were significantly more durable with escalation of treatment, but it is noteworthy that adverse events such as grade 3 to 4 febrile neutropenia were more common in the eBEACOPP group (23.9% vs 1.1%), and OS was not significantly different, despite a trend favouring eBEACOPP: 89% vs 96%, P = 0.062.15 The international Response-adapted therapy in Hodgkin lymphoma (RATHL) and Italian Gruppo Italiano Terapie Innovative nei Linfomi (GITIL) studies have both assessed treatment reduction following 2 cycles of ABVD. In RATHL, patients with a negative iPET were randomised to either continue ABVD or de-escalate to Doxorubicin, vinblastine, bleomycin (AVD).16 Three-year PFS was 85.7% and 84.4% for the ABVD and AVD groups, respectively, with a difference of 1.6 percentage points (95% CI, −3.2 to 5.3). Overall survival after 3 years was also similar (97.2% for ABVD and 97.6% for AVD). Thus, after a negative iPET, the omission of bleomycin did not lead to higher rates of recurrence or mortality, a finding that was replicated across all subgroups including stage, disease bulk, and IPS. Furthermore, there were lower rates of pulmonary adverse events in the AVD subgroup, and significantly better preservation of lung diffusion capacity up to 1 year after treatment. Whilst indicating that iPET can be used to safely de-escalate therapy, the RATHL study also echoed an observation of other prospective trials: that the negative predictive value of iPET is less than that reported in earlier retrospective studies. This finding was more pronounced in more advanced cases, with a 3-year PFS of 79.6% in stage IV disease, compared with 90% in stage II. It is interesting to compare these observations with the LYSA AHL2011 study, in which patients with a negative iPET after 2 cycles of eBEACOPP went on to complete 4 cycles of ABVD.17 Two-year PFS was 92% in the de-escalated ABVD group, and 94% for those remaining on eBEACOPP, indicating that this represents another safe de-escalation strategy. It also suggests that iPET is more reliable in patients treated with more intensive chemotherapy. Hence, the need to improve the negative predictive value of iPET in advanced disease, as highlighted by the RATHL study, could be met by using more intensive regimes from the outset in the especially high-risk group. Escalation of treatment to eBEACOPP following 2 cycles of ABVD has been evaluated by the GITIL 0607 study, the US Intergroup Study and the RATHL trial.16, 18, 19 The results of these studies are strikingly similar, reporting a 3-year PFS of 65% to 70% in the patients with a positive iPET. It is important to note when interpreting these trials that there was no control arm, owing to the previously reported poor prognosis for iPET-positive patients who continued to receive ABVD. However, by comparison with earlier retrospective studies, which reported durable remissions in less than 30% of this particular group, it is plausible to conclude that escalation of therapy to eBEACOPP is an effective measure, a conclusion supported by the results in early stage disease in the H10 study. This still leaves a third of iPET-positive patients needing further treatment for relapse within 3 years, so clearly there is room for improvement, especially among the small number with an iPET score of 5, where the failure rate after BEACOPP was 50% in RATHL. More intensive regimens have been considered for this group, as exemplified in the Italian Lymphoma Group HD0801 phase II trial, in which patients with a positive iPET after 2 cycles of ABVD went on to receive 4 cycles of salvage therapy with ifosfamide, gemcitabine, and vinorelbine followed by myeloablative chemotherapy and autologous stem cell transplantation.20 The 2-year PFS was 76% for iPET-positive patients, although in this study the iPET-positive group included those with a score of 3 on the 5-point Deauville scale, which may have improved the PFS figure significantly. With the dangers of treatment-related toxicity and the association between positive iPET and worse outcomes, a strong case can be made for response-adapted therapy, both de-escalating treatment in those responding well to limit long-term toxicity and intensifying therapy in nonresponders to try and achieve durable remissions. The data on early disease suggest that a decision can be taken based on the outcome of iPET for the safe omission of consolidation radiotherapy or appropriate escalation of chemotherapy. In advanced disease, iPET must be interpreted carefully, since its negative predictive value diminishes in those with the worst disease and those treated with less intensive chemotherapy. There is an argument to treat patients with very bad baseline prognostic features with more intensive therapy prior to iPET. The studies performed have identified safe methods for the de-escalation of both ABVD and eBEACOPP, as well as effective means of intensifying therapy to achieve better outcomes following positive iPET. Finding the right balance between efficacy and toxicity remains a significant challenge in treating HL. Although not perfect, iPET offers an early assessment of response to treatment, which allows a tailored approach to be taken with each patient, and is rapidly becoming the standard of care. In the future, gene expression and biomarker analysis may prove useful for further stratifying patients, and novel agents such as brentuximab vedotin and PD-1/PDL-1 checkpoint inhibitors can be integrated into the response-adapted approach to further individualise therapy and improve outcomes. Dr Johnson needs to declare that he has undertaken remunerated consulting for Takeda and Bristol-Myers Squibb. Dr Broadfoot has no conflicts to declare.