Physician frontline treatment preferences for stage III/IV classic Hodgkin lymphoma: the real-world US CONNECT study

Hodgkin lymphoma is uncommon, with an incidence of approximately 8500 new cases per year in the USA, of which 95% are classic Hodgkin lymphoma (cHL) [1,2]. Approximately 40–50% of adults with newly diagnosed cHL have stage III/IV disease [1]. For stage III/IV cHL, ABVD (doxorubicin, bleomycin, vinblastine and dacarbazine) has been a standard frontline (1L) therapy for several decades. However, 30% of patients will have refractory disease or relapse following treatment with ABVD, requiring subsequent salvage therapy, potentially followed by high-dose therapy with autologous stem cell rescue [3–5].

While many patients with cHL who receive treatment can experience long-term survival, treatment selection is important, as the lifetime survivorship burden can be high with 1L therapies such as ABVD and escalated BEACOPP (escalated dosing regimen of bleomycin, etoposide, doxorubicin, cyclophosphamide, vincristine, procarbazine and prednisone). Pulmonary toxicity, infertility, second cancers and cardiovascular effects are concerns following treatment with ABVD and escalated BEACOPP regimens [6–13]. In an effort to maintain disease control while limiting toxicity, response-adapted therapies, such as those employed in the RATHL and SWOG SO816 trials, refine treatment based on results from an interim positron emission tomography (PET) scan following two cycles of ABVD, with early escalation of therapy to BEACOPP for patients with a positive interim PET scan [14–20].

In the USA, 1L therapy with escalated BEACOPP is rarely used as it is associated with both increased short- and long-term toxicity compared with ABVD [14,15,19,21,22]. However, 10-year results from the HD9 German Hodgkin Study Group study do show improved overall survival (OS) in patients treated with escalated BEACOPP compared with alternating cycles of ABVD and COPP (cyclophosphamide, vincristine, procarbazine and prednisone) [23]. Results from a systematic literature review and meta-analysis that included data from five randomized controlled trials comparing outcomes for 1L chemotherapy regimens that included two or more cycles of escalated BEACOPP with those containing four or more cycles of ABVD for early unfavorable-stage or advanced-stage HL also showed an OS advantage with escalated BEACOPP versus ABVD [24]. In contrast, no OS advantages have been reported for response-adapted therapies compared with traditional ABVD [17,19,20,22,25,26]. Of note, results from a preference survey administered to physicians who treated HL in France, Germany and the UK found that physicians valued OS most when selecting a 1L treatment [27].

Brentuximab vedotin (Adcetris^®) in combination with doxorubicin, vinblastine and dacarbazine (A+AVD) is a 1L treatment option for patients with newly diagnosed stage III/IV cHL. In ECHELON-1, at approximately 6 years of follow-up, patients newly diagnosed with stage III/IV cHL randomized to A+AVD experienced a 41% reduced risk of death compared with patients randomized to ABVD (hazard ratio [HR]: 0.59; 95% CI: 0.40–0.88; p = 0.009) and a 32% reduction in the risk of disease progression or death (HR: 0.68; 95% CI: 0.53–0.86).

Given the importance of selecting a 1L treatment for patients with stage III/IV cHL, we surveyed USA-based physicians as part of the CONNECT (Classic Hodgkin Lymphoma: Real-World Observations From Physicians, Patients and Caregivers on the Disease and Its Treatment) study—the first real-world survey in cHL to include patients, physicians and caregivers—to better understand physician preferences for 1L treatment regimens for adults with newly diagnosed stage III/IV cHL and the factors that influence treatment decision-making.

Materials & methods

Study design

The CONNECT physician survey was an anonymous online survey developed based on areas of interest with regards to 1L cHL treatment preferences aligned with consensus guidelines and in collaboration with expert clinicians. It was administered from 19 October 2020 through to 16 November 2020. Participating physicians were blinded to the study sponsor, and the study sponsor was blinded to the survey participants. The survey was designed to be completed in 30 minutes. All data from the physicians were collected anonymously via a secure online data collection platform. Physicians were compensated up to US$170 for their time.

Several steps were taken to ensure data quality. Skip logic was implemented such that respondents were asked only relevant questions. Range checks were included to minimize erroneous responses that were outside the valid range and inconsistent with previous responses. A multilevel review of test links was also performed to ensure the program's accuracy.

The quality of data achieved via the survey was evaluated. Responses obtained from participants who finished the survey substantially faster than the average response time or provided multiple unrealistic, internally inconsistent responses were not included in the analysis. Missing data were minimal, as respondents did not have the option to not respond for most questions. Any responses that were unknown were not utilized in statistical analyses; this includes missing values (if any), which were also treated as unknown.

Study participants

Physicians from the USA were recruited via email using multiple, existing, non-mutually exclusive online panels of healthcare providers. Because of this, the exact number of unique physicians invited is unknown, and response rates were not calculated. The panels seek to be representative of the USA physician population and leverage multiple sources to recruit physicians to increase reach and capacity, improve consistency and minimize bias. Physicians were screened at the beginning of the survey to ensure that they met inclusion criteria.

Eligible physicians included those who self-identified as medical oncologists, hematologist/oncologists, or hematologists with at least 2 years of medical practice experience who had treated at least one adult (aged at least 18 years) with stage III/IV cHL and at least one adult with cHL in the 1L setting within the past 12 months. Physicians practicing in Maine or Vermont were excluded, as Sunshine Act laws in these states could unblind study participants to the study sponsor.

Statistical analyses

Data were summarized as mean and standard deviation (SD) or median and range. Categorical data were reported as individual totals or percentages. Non-mutually exclusive data were reported as a number and percentage of total sample size.

Consent & compliance

This study was conducted in full compliance with the Guidelines for Good Pharmacoepidemiology Practices published by the International Society for Pharmacoepidemiology [28] and the laws and regulations of the USA, the country where this research was conducted. This study was reviewed by the New England IRB and received exemption status.

Results

Physician demographics

Overall, 301 physicians participated in the survey. The majority of physicians (80%) self-identified as having a primary medical specialty in hematology/oncology; 20% self-identified as having a primary medical specialty in medical oncology. The majority of physicians had a practice in the community setting (62%) versus academic setting (38%). 34% were located in the South, 26% in the Northeast, 21% in the West and 20% in the Midwest as defined by the United States Census Bureau [29]. Participating physicians practiced for a mean of 16 years and reported spending a mean of 90% of their time in direct patient care.

In the 12 months prior to survey participation, 16% of physicians reported seeing more than 100 adult patients with active cHL or cHL survivors (high volume), 37% reported seeing 31 to 100 patients with active cHL or cHL survivors (medium volume), and 47% reported seeing 30 or fewer patients with active cHL or cHL survivors (low volume). Physicians reported caring for a median (range) of 16 (7–40) adults with active cHL and 15 (8–40) cHL survivors in the prior 12 months. Most of their adult patients with cHL had stage III/IV disease (59%) and were aged 18–39 years (41%).

Overall frontline cHL treatment considerations

Physicians reported clinical trial efficacy and safety data and official guideline recommendations as their most important considerations (ranked 1 or 2) when selecting 1L cHL treatments (Figure 1A). The following attributes were rated by more than 80% of physicians as having a moderate-to-high or essential impact on their decision-making when considering 1L cHL treatments for stage III/IV disease: OS (91%), long-term progression-free survival (PFS; 86%), curative potential/higher cure rates (85%) and complete response rate (82%; Figure 1B). Guidelines that were considered by 50% or more of participating physicians when treating cHL were those from the National Comprehensive Cancer Network^® (NCCN; 88%) and institution-specific formularies/clinical pathways (50%; Figure 1C).

Fewer than 10% of physicians assigned a rank of 1 or 2 to the following 1L cHL treatment attributes: patient personal goals, treatment costs and financial support programs (Figure 1A). Despite this, nearly all physicians reported taking patient preference into account to some degree when selecting treatment. More than 90% of physicians reported discussing treatment pros and cons, and 77% reported discussing side effect trade-offs with patients.

The impact of patient characteristics on treatment decisions was also evaluated. Physicians were asked to rank the top three most important patient characteristics they considered when evaluating a 1L cHL treatment regimen for patients with stage III/IV cHL from the following list: age, sex, comorbidities (e.g., respiratory, cardiovascular disease), Deauville 5-point scale (5PS) score, standardized uptake value, presence/absence of B symptoms, presence/absence of a bulky mass, disease stage, fitness/frailty (measured using the Eastern Cooperative Oncology Group performance score), International Prognostic Score and patient goals/preferences (e.g., preferred treatment period). The patient characteristics ranked 1 or 2 in importance by more than 20% of physicians were disease stage (50%), fitness/frailty (35%), comorbidities (28%) and International Prognostic Score (24%; Figure 2).

Frontline treatment preferences for stage III/IV cHL

Physicians were asked to choose their first-choice treatment for various patient types (bulky mediastinal disease, stage III, stage IV, perceived high risk of relapse, elderly and younger, more fit) with stage III/IV cHL from the following three treatments commonly administered in the USA: A+AVD, ABVD (without PET adaptation), or PET-adapted ABVD (Figure 3). Across these patient types, 37–50% of physicians selected A+AVD, 20–26% selected ABVD, and 25–38% selected PET-adapted ABVD as their first-choice treatment regimen.

Newly diagnosed stage III/IV cHL patient profiles

Physicians were presented several hypothetical patient profiles reflecting different types of cHL patients and asked questions about which 1L treatment options they would administer with or without radiotherapy for each of these patient profiles from the following eight options: ABVD, BEACOPP, A+AVD, Stanford V, brentuximab vedotin monotherapy, AVD, PET-adapted ABVD and other (Figure 4). Although 1L treatments varied across the patient types, A+AVD or ABVD with and without PET adaptation were the most common 1L treatments selected, except for administration to the older, unfit patient with stage III/IV cHL, illustrated by Patient 4, for whom a brentuximab vedotin-based therapy was preferred by 51% of participating physicians (A+AVD, 30%; brentuximab vedotin monotherapy, 21%; Figure 4A). Across these four patient profiles, BEACOPP and Stanford V were preferred by ≤5% of participating physicians. Across all hypothetical patient profiles, OS rate was the top reason cited by physicians for selecting a 1L treatment (Figure 4B). When we evaluated reasons for choosing treatments, regardless of patient profile, patient age, fitness/frailty, comorbidities and quality of life were more commonly cited as reasons for choosing a less intensive treatment regimen (e.g., brentuximab vedotin monotherapy or AVD [doxorubicin, vinblastine and dacarbazine]; Figure 5).

Discussion

In this USA-based study, physicians consistently selected OS as the top treatment attribute considered when making 1L treatment decisions for previously untreated patients with advanced-stage cHL, including treatment decisions for specific patient archetypes with newly diagnosed stage III/IV cHL. While this is the first study to survey physicians on 1L cHL treatment preferences in the USA, our results mirror a preference survey administered to patients with HL and physicians who treated HL in France, Germany and the UK, which also found that physicians preferred treatment with higher OS and PFS outcomes [27]. Results of the 2018 survey also highlighted the importance of patient characteristics in physicians' treatment selection.

In the current survey, although treatment preferences for patients with newly diagnosed stage III/IV cHL varied based on patient characteristics, including the presence of bulky mediastinal disease, disease stage, perceived risk of relapse, age and comorbidities, the most common reason cited by physicians for choosing a particular therapy continued to be OS. Age, comorbidities, side effects and quality of life played a larger role in the selection of less intensive therapies.

These results highlight the importance of efficacy results, particularly OS, when selecting 1L therapies for patients with stage III/IV cHL and the importance of choosing the most efficacious regimen that decreases downstream clinical and economic burden. 6-year OS results from the ECHELON-1 trial, although published after this survey was conducted, are the first to show an OS benefit with A+AVD compared with ABVD alone in patients with newly diagnosed stage III/IV cHL [30]. The demonstrated OS benefit with 1L A+AVD compared with ABVD was observed across prespecified subgroups, including both stage III and IV disease, and was independent of PET status at the end of two cycles of therapy. Furthermore, results from an oncology simulation model support improved OS and PFS over 10 years for patients with stage III/IV cHL with increased A+AVD and decreased ABVD utilization in the USA [31].

Improved OS at 10 years was reported with escalated BEACOPP in the HD9 Study by the German Hodgkin Study Group compared with alternating cycles of ABVD and COPP [23]. Additionally, improved OS was reported from a meta-analysis of five randomized controlled trials comparing 1L chemotherapy regimens consisting of at least two cycles of BEACOPP compared with regimens containing at least four cycles of ABVD for the treatment of early unfavorable-stage or advanced-stage HL [24]. However, 1L therapy with escalated BEACOPP is rarely used in the USA as it is associated with both increased short- and long-term toxicity compared with ABVD [14,15,19,21,22].

For the older, unfit patient archetype, the majority of physicians who participated in CONNECT chose brentuximab vedotin either alone or in combination with AVD for 1L treatment. Although OS was still the most common reason cited for selecting a treatment for the older patient archetype, fitness/frailty, comorbidities and age were also cited. Results from ECHELON-1 show that at a median of 61 months, a numerically higher proportion of patients aged ≥60 years achieved 5-year PFS with A+AVD versus ABVD (67.1 vs 61.6%; HR: 0.820 [95% CI: 0.494–1.362]; p = 0.443), although survival rates were similar. In ECHELON-1, regardless of treatment received, a higher incidence of adverse events was reported in older versus younger patients, emphasizing the potential need for alternative regimens that are well tolerated and provide robust outcomes. Results from a phase II study that enrolled 42 patients aged ≥60 years (median 69 years) showed overall response and complete remission rates of 95 and 90%, respectively, following treatment with two lead-in doses of brentuximab vedotin monotherapy, followed by six cycles of AVD. Patients who responded to this regimen received four consolidative brentuximab vedotin doses following AVD [32]. Safety outcomes for this phase II study compare favorably with those from other studies, with minimal risk of grade 3 neuropathy reported.

In younger patients diagnosed with cHL, similar percentages of physicians selected PET-adapted ABVD, A+AVD, and ABVD as 1L therapy. Reasons cited for selecting these regimens were primarily OS rate and PFS rate. 6-year OS results from ECHELON-1 specifically for the subgroup of patients aged < 45 years, not available at the time of the survey's deployment, showed an OS advantage for A+AVD compared with ABVD (HR: 0.44; 95% CI: 0.20–0.99) [30].

Approximately 30% of physicians participating in the CONNECT survey preferred PET-adapted ABVD as 1L therapy for patients with stage III/IV cHL; although PET-adapted ABVD is guideline recommended for these patients, there are challenges with this approach [33,34]. As part of a PET-adapted approach, guidelines recommend an interim PET scan after two cycles of ABVD to inform treatment adaptation [33–35]. However, results from this CONNECT physician survey on PET scan use (published separately) [36] and from real-world retrospective analyses of US Oncology Network health records [37] and Flatiron Health data [38] suggest that interim PET scans after cycle 2 are not universally obtained. Furthermore, when interim PET scans are obtained, Deauville 5PS scores, recommended for PET scans, are not consistently provided; additionally, when Deauville 5PS scores are provided, the definitions used by physicians to determine what is a positive versus negative score vary [36]. Despite these findings, de-escalation to AVD was observed, and escalation to BEACOPP was rare, suggesting inconsistent use of the PET-adapted approach [37,39]. These results highlight the need to educate physicians on PET-adapted treatment approaches to improve the quality of image reporting and interpretation, and the need to alter therapy based on imaging results [36–38].

In addition to safety and efficacy, value should also be considered in the selection of a 1L treatment for cHL. Results from this analysis show that treatment costs were generally not considered by physicians (i.e., fewer than 5% ranked them most important) when choosing therapies commonly administered 1L for stage III/IV cHL in the USA (i.e., A+AVD, PET-adapted ABVD, and ABVD) but were mentioned by 10% of physicians choosing regimens less commonly administered in the USA (i.e., BEACOPP, Stanford V). Although previously conducted economic analyses suggest varying levels of cost–effectiveness with 1L A+AVD compared with ABVD, these analyses did not incorporate OS outcomes [40,41]. Results from an analysis evaluating productivity losses that did incorporate the ECHELON-1 OS data suggest decreases in productivity losses by 14–32% in the USA over 10 years with increasing A+AVD use from 0 to 80% [42]. Additionally, the failure of 1L therapy is associated with significant economic burden, incurring an estimated projected economic impact of over US$530,000 per patient for patients receiving two or more lines of treatment over 5 years [43]. While this estimate includes the burden of additional therapy, it is also important to consider the increased costs for healthcare resource utilization, including treatment-related supportive care and SCT. The use of SCT, a potentially curative treatment for patients after 1L failure, is associated with considerable healthcare resource utilization and economic burden. Total all-cause costs for patients with hematologic malignancies were shown to be significantly higher for autologous and allogeneic SCT recipients versus non-SCT recipients (US$540,194 vs $182,846 and US$929,799 vs $190,776; p < 0.001) over an 18-month observation period [44].

Limitations

Results from this study may not be generalizable to all physicians who treat patients with cHL due to the opt-in group of survey participants who were already part of established research panels. Additionally, physicians participating in this survey were only from the USA. Drivers of physician decision-making may vary across other geographies, cultures and funding environments; this is particularly relevant for geographic areas using escalated BEACOPP as a standard therapy as escalated BEACOPP is rarely used in the USA.

Recall bias may also affect outcomes, as physicians were asked to recall practice patterns over the previous 12 months. Furthermore, physicians counted all patients with active disease; this tally may have included patients not newly diagnosed as well as patients experiencing relapse. Of note, we did not probe participating physicians on financial incentives for selecting a 1L therapy; therefore, the impact of financial interests on treatment selection is unknown. While we acknowledge these limitations, with more than 300 physicians contributing to the survey, the CONNECT study, to our knowledge, is the largest survey of physicians who treat patients with cHL. The survey was administered prior to publication of the ECHELON-1 6-year follow-up data, which demonstrated an OS and PFS benefit with A+AVD compared with ABVD.

Conclusion

Overall survival, the most important end point in oncology, continues to be a key outcome of interest for physicians in the USA treating patients with cHL and is one of the most prominent factors physicians consider when selecting a 1L treatment for newly diagnosed patients with stage III/IV cHL. Although physicians' treatment preferences in the USA for patients with stage III/IV cHL varied based on patient characteristics, including the presence of bulky mediastinal disease, disease stage, perceived risk of relapse, age and comorbidities, OS continued to be the top reason cited by surveyed physicians for selecting a specific 1L treatment regimen for stage III/IV cHL.