Results of a UK National Cancer Research Institute Phase II study of brentuximab vedotin using a response‐adapted design in the first‐line treatment of patients with classical Hodgkin lymphoma unsuitable for chemotherapy due to age, frailty or comorbidity (BREVITY)

Standard treatment for classical Hodgkin lymphoma (cHL) is poorly tolerated in older patients and results disappointing. We assessed safety and efficacy of brentuximab vedotin (BV), in previously untreated patients with cHL unfit for standard treatment due to age, frailty or comorbidity. The primary outcome was complete metabolic response (CMR) by positron emission tomography/computed tomography after four BV cycles (PET4). The secondary outcomes included progression‐free survival (PFS), overall survival (OS), and toxicity. In all, 35 patients with a median age of 77 years and median total Cumulative Illness Rating Scale for Geriatrics (CIRS‐G) score of 6 were evaluable for toxicity and 31 for response. A median of four cycles were given (range one–16). In all, 14 patients required dose reduction due to toxicity and 11 patients stopped treatment due to adverse events (AEs). A total of 716 AEs were reported, of which 626 (88%) were Grade 1/2 and 27 (77%) patients had at least one AE Grade ≥3. At PET4, CMR was 25·8% [95% confidence interval (CI) 13·7–42.2%] and objective response rate 83·9% (95% CI 63·7–90·8%). Median PFS was 7·3 months (95% CI 5·2–9·0), and OS 19·5 months. Our results suggest that BV monotherapy is tolerable but suboptimal in the front‐line therapy of elderly or comorbid patients with cHL. Combining BV with other agents may be more effective.


Introduction
Approximately 2100 people are diagnosed with classical Hodgkin lymphoma (cHL) in the UK each year. There are two peaks in the age-specific incidence; one in young adults aged 20-24 years and another in older people aged 75-79 years who form 13% of new diagnoses. 1 HL in younger people is treated with front-line chemotherapy usually ABVD (doxorubicin, bleomycin, vinblastine and dacarbazine) or BEACOPP (bleomycin, etoposide, doxorubicin, cyclophosphamide, vincristine, procarbazine and prednisolone) with or without consolidation radiotherapy. According to stage and other risk factors the 3-year progression-free survival (PFS) varies between 67% and 95%. 2,3 However, outcomes are significantly worse for people aged >60 years. One study reported a 5-year PFS of 46% and 58% for advanced and early stages respectively, with age ≥70 years and inability to perform the activities of daily living (ADL) being associated with the worst outcomes. 4 In another series, 23% of patients with HL aged >65 years died of complications related to ABVD. 5 It is hypothesised that older individuals have a worse outcome due to more aggressive histology or differences in fundamental disease biology. In addition, standard chemotherapy is often poorly tolerated leading to dose reductions/delays and reduced dose intensity. Moreover co-existing heart disease, lung disease or general frailty may render it too dangerous in the opinion of the treating physician to use standard chemotherapy such as ABVD or BEACOPP, both of which contain agents known to be specifically toxic to the heart (doxorubicin) or lungs (bleomycin). Consequently, better tolerated but less effective therapies are employed.
Brentuximab vedotin (BV) is a CD30 targeted antibodydrug conjugate (ADC) composed of the anti-CD30 monoclonal antibody cAC10 and a potent anti-microtubule drug, monomethyl auristatin E (MMAE). cAC10 binds to the CD30 antigen, which has very low expression on normal cells, but is consistently expressed on Hodgkin Reed-Sternberg cells. Phase I 6 and II 7 data in the relapsed/refractory HL population suggest that BV is highly effective with a very manageable toxicity profile and no known cardiac or pulmonary toxicity when given as a single agent. These characteristics open up the possibility of using the drug to treat patients with newly diagnosed HL where standard doxorubicin and bleomycin containing chemotherapy is considered inappropriate because of co-morbidity or frailty.
Levels of soluble CD30 (sCD30) present in the serum of newly diagnosed patients with HL treated with ABVD (or similar), 8 and relapsed patients with HL treated with BV, 9 have been shown to be prognostic, with lower baseline levels associated with better outcomes. Similar findings have been reported for thymus and activation-regulated chemokine (TARC, also known as CCL17) in newly diagnosed patients treated with chemoradiotherapy, 10 and relapsed patients treated with BV, chemotherapy and haemopoietic stem cell transplant. 11 In BREVITY, we aimed to test BV monotherapy in a population of patients with previously untreated cHL who were elderly, frail or had significant comorbidities at any age and for whom conventional chemotherapy was not considered a safe option by their treating physician. Clinical outcomes were correlated with baseline prognostic and comorbidity indices, and serial fluorodeoxyglucose-positron emission tomography/ computed tomography (FDG-PET/CT) responses.

Patients
Patients were recruited at 13 UK haemato-oncology centres. Previously untreated patients with stages II (with B symptoms and/or mediastinal bulk), III and IV cHL were eligible if considered unfit for standard chemotherapy. This was assessed by the treating investigator, but had to include at least one of the following factors: Impaired cardiac function defined either by an ejection fraction of <50% assessed by echocardiogram or nuclear medicine [multigated acquisition (MUGA)] scan.
Left ventricular ejection fraction of ≥50% measured by echocardiography or MUGA, but in the presence of significant comorbidities or cardiac risk factors such as diabetes mellitus, hypertension, peripheral vascular disease, ischaemic heart disease, previous myocardial infarction, obesity, stroke or transient ischaemic attacks (TIA) that make anthracycline-containing chemotherapy inadvisable as determined by the investigator.
Heart failure clinically determined by the presence of New York Heart Association (NYHA) heart failure grade II and III due to a cause other than HL.
Impaired respiratory function with carbon monoxide diffusion capacity (DLCO) and/or forced vital capacity/ forced expiratory volume in 1 s (FVC/FEV1) ratio <75% of predicted due to a cause other than HL.
Eastern Cooperative Oncology Group Performance Status (ECOG PS) score 1-3 in patients aged ≥60 years.
Further details are provided in the Online Supplement.
Study design, end-points and treatment schedule BREVITY is a Phase II, single-arm, adaptive-response study to investigate the level of activity and tolerability of BV as a single agent. The primary outcome was complete metabolic response (CMR) after four cycles of BV defined as a Deauville Score of 1, 2 or 3 by 18 F-FDG-PET/CT scan, 12 from which the CMR rate (CMRR) was calculated. Secondary outcomes included overall response (OR) as per the Lugano Classification, 13 defined as complete plus partial metabolic response (CMR + PMR) after two and four cycles of BV (PET2, PET4), from which the OR rate (ORR) was calculated. The PET2 results were blinded to the investigators to prevent early discontinuation in cases of pseudoprogression or other indeterminate responses. All scans were performed on approved PET-CT scanners with quality assurance co-ordinated by the UK PET Core Laboratory at St Thomas' Hospital, London, UK. 14 Survival outcomes included PFS, overall survival (OS) and the correlation between PET4 and survival. Tolerability of BV was defined in terms of absence of toxicities related to BV quantified by the Common Terminology Criteria for Adverse Events (CTCAE) version 4 criteria, as well as dose intensity defined as the total dose delivered to each patient as a proportion of the planned protocol dose.
BV was administered at an initial dose of 1Á8 mg/kg every 3 weeks as a 30-min outpatient intravenous (IV) infusion. The IV contrast-enhanced CT scans were performed at baseline and following cycles four, eight, 12, 16 or at study exit. Patients in response at PET 4 could continue with treatment providing no evidence of progressive disease was seen at any of these time points. Details of the treatment schedules, permitted dose reductions, assessments, study end-points and molecular methods are reported in the Online Supplement.

Follow-up and statistical analysis
Patients were followed-up for up to 3 years from day 1 of cycle one. A Simons two-stage minimax design determined 30 evaluable patients were required, with alpha at 20% and power at 90%. Registered patients subsequently found to have been ineligible at baseline or not starting treatment were to be replaced, although would still contribute toxicity data. Stage 1 recruited 20 eligible patients. Initially an unacceptable CMRR (p0) was set at 40%, below which BV would not be recommended for further investigation. Following agreement amongst the investigators that patients in PMR were deriving significant clinical benefit from BV monotherapy the protocol was amended in October 2014 to allow them to continue on BV. Recruitment into the study was much faster than expected. At the point where stage 1 recruitment was completed there were seven outstanding responses awaited from which three CMR responses would be required to reach the eight required overall to continue to stage 2. The Trial Steering Committee (TSC) reviewed all available data and concluded that as patients were deriving significant clinical benefit without excessive toxicities that recruitment should continue to the second stage without a pause to await the seven outstanding responses. An acceptable response rate (p1) was set at 60%, above which BV would definitely be worthy of further investigation. Response rates were calculated and reported as proportions, and a 95% confidence interval (CI) was constructed using Wilson's estimates due to its increased accuracy with small sample sizes. Time-to-event outcomes were analysed using the Kaplan-Meier method and the median, 12 and 24 month estimates are reported. Correlation of response at PET4 with PFS was conducted using a Cox regression model. Planned dose intensity per cycle was calculated for each patient using the protocol specified dose per kg and 21-day cycle length. Average dose intensity administered per patient was calculated using the actual dose given and the actual cycle length (accounting for delays). The relative dose intensity is defined as the actual dose intensity as a proportion of the planned dose intensity.

Patients
Recruitment to BREVITY completed ahead of the planned accrual date, with a total of 38 patients recruited between 14 February 2014 and 20 October 2015, with a median follow-up of 3 years. Three patients did not receive BV (two ineligible, one failed to start treatment). Of the 35 patients who received BV a further four patients were found to be ineligible upon subsequent review. Following agreement from the TSC, these seven patients were replaced, although it was agreed that every patient who received BV would be evaluated for toxicity. In total 31 patients were eligible and hence evaluable for efficacy ( Figure S1). All patients were followed-up. Baseline characteristics are summarised in Table I, including the Cumulative Illness Rating Scale for Geriatrics (CIRS-G) data.
The 35 patients reported a total of 716 adverse events [AEs; median (IQR; range) 13 (6,25; 1-94), of which the majority (88%) were CTCAE Grade 1 or 2. Only two patients reported no AEs related to BV. In all, 31 evaluable patients reported a total of 246 events related to BV [median (IQR; range) 5 (3,8; 1-20)]; 84% of these were Grade 1 or 2. In all, 18 patients had at least one related AE of Grade ≥3 and 14 serious AEs (SAEs) were reported, 27 of which were deemed to be related to BV. Three SAEs were fatal and of these two were unrelated to BV and one (sepsis) was considered related. Figure 1 shows all adverse events experienced by at least 10% of patients, according to relationship to BV.

Response
A total of 31 patients were deemed evaluable for the primary end-point and 25Á8% (95% CI 13Á7-43Á2%) achieved CMR after four cycles. This was lower than the 40% required by the initial trial design to indicate sufficient activity of BV in this patient population to warrant further investigation.
Despite the low CMR rate, the ORR (CMR + PMR) at PET4 was 83Á9% (95% CI 63Á7-90Á8%). In addition to the PET scan after four cycles, 25 patients had a PET scan after two cycles and seven had a PET after 16 cycles. The ORR after two cycles was 80Á6% (95% CI 63Á7-90Á8%). A breakdown of responses is shown in Table II. A total of 24 patients underwent both PET2 and PET4 scans; of these, response for 20 patients did not change between the two scans (six CMR, 14 PMR) and one patient improved their response from PMR to CMR.
Two patients with PMR at PET4 were reported to have CMR at PET2, when this result was unblinded. Both patients were deemed to have achieved PMR for the purposes of the primary end-point of the study by the TSC. Both patients' scans showed subtle changes between PET2 and PET4, and they both underwent unequivocal clinico-radiological progression at 7Á3 and 8Á8 months respectively.
Progression-free survival and overall survival. To date 28 of 31 evaluable patients have progressed. Of three patients who were progression free at the end of the trial, one patient withdrew after 6Á7 months and the other two were followed up for a minimum of 29 months. The median PFS time was 7Á3 months (95% CI 5Á2-9Á0). At 12 and 24 months, 13Á7% (95% CI 4Á3-28Á4) and 6Á9% (95% CI 1Á2-19Á6) of patients were progression free (Fig 2). The median OS was 19Á5 months (95% CI 12Á6-not reached). Survival estimates at 12 and 24 months were 73Á4% (95% CI 53Á7-85Á7) and 42Á0% (95% CI 24Á1-58Á8) respectively. Kaplan-Meier curves exploring the relationship between response and both PFS and OS are shown in Fig 3. This analysis is exploratory due to the small numbers of patients, and the survival estimates are biased upwards as only those patients who reached the assessment points could be included.

sCD30 and TARC
An exploratory analysis of sCD30 and TARC was performed on serum from 10 patients. The TARC and CD30 results were obtained for all 37 samples analysed; for eight samples the CD30 assay fell below the level of detection. Baseline levels of CD30 and TARC varied dramatically between patients; mean (range) baseline CD30 and TARC levels were 347Á4 (30Á3-1720Á9) pg/ml and 45 064 (391Á4-181 681) pg/ml respectively. All 10 patients in this analysis had CMR or PMR at PET4, and tended to show an increase in CD30 levels from baseline measurements. An initial decrease in TARC between baseline and cycle three was observed for nine of the 10 patients; these analyses are presented graphically in

Discussion
In BREVITY, BV monotherapy produced a high ORR and modest CMR rate consistent with published data in the relapse setting; 7 however, the success criterion for the primary outcome was not achieved. Toxicity was common and mainly mild to moderate in severity; the mean dose intensity approaching 90% indicates that this is a tolerable therapy that can be delivered in the outpatient setting.
The PFS and OS were short; published studies of chemotherapy resulted in higher PFS and OS rates, but the BREVITY population specifically included patients deemed unsuitable for chemotherapy. 4 Biomarker analysis of sCD30 and TARC was limited, but corroborated previously reported decreases in TARC levels following initial treatment 15 and at PET2. 16 Previously, TARC levels have been shown to remain high in non-responsive patients 15 ; the initially high levels measured in the present study may be linked to the poor survival outcomes of these patients. Similarly, sCD30 tended to increase during treatment, and elevated levels of sCD30 have been linked with poor outcomes and disease relapse. 17 Exploratory proteomic analysis is underway and will be reported separately.
Other researchers have conducted similar studies with BV in elderly/unfit patients with HL, both as monotherapy 18 and in combination with the cytotoxic agents bendamustine 19 or dacarbazine. 20 Ferero-Torres et al. 18 treated 27 patients with HL with BV monotherapy. The median age was 78 years and 67% were impaired in at least one ADL. The ORR was 92%, with 73% achieving complete remission. However, similar to The results of BREVITY and similar studies suggest that BV monotherapy is suboptimal in the front-line therapy setting for elderly and/or comorbid patients with cHL. Although disappointing, outcomes are not dissimilar to those reported for chemotherapy in similar populations of elderly or frail patients with HL, suggesting that HL in this demographic remains a challenging disease. Nevertheless, whilst BV monotherapy failed to have a substantial impact on outcomes for these patients, treatment was tolerable and it may be feasible to combine BV with cytotoxic chemotherapy or other novel agents to achieve better results. Results of BVchemotherapy combination studies look promising, and BV has also been combined with the immune checkpoint inhibitor nivolumab in the relapsed/refractory setting with encouraging early results. 21,22 Funding This trial is funded by Bloodwise under the Trials Acceleration Programme (TAP). An unrestricted educational grant was provided to support the trial and adjunctive science by Millennium -The Takeda Oncology Company. Brentuximab vedotin was provided free of charge by Millennium -

Supporting Information
Additional supporting information may be found online in the Supporting Information section at the end of the article.