Zanubrutinib for treating marginal zone lymphoma after anti-CD20-based treatment

The clinical-effectiveness evidence for zanubrutinib came from 2 international, multi-centre, single-arm, open-label trials: MAGNOLIA (phase 2; n=68) and AU‑003 (phase 2 part of a phase 1 and 2 study; n=20). The trials included people aged at least 18 years with relapsed or refractory MZL, who had had at least 1 previous line of treatment. Only MAGNOLIA included some people based in the UK. The primary outcome was best overall response assessed by an independent review committee. The company used the secondary outcomes of progression-free and overall survival in its economic model. The committee noted that the average age of people in the trials was less than 70 years. This was younger than people likely to be seen in the NHS, where people are usually diagnosed aged around 75 years (see section 3.1). The committee also noted that the median progression-free and overall survival had not been reached in either trial. It considered that the people in the trial may not be fully representative of people likely to have zanubrutinib in the NHS. It acknowledged that with no direct comparative clinical evidence, it was hard to interpret the trial results, and the immature survival trial data increased the uncertainty. The committee concluded that these areas of uncertainty would be considered in its decision making.

The company used data from a UK-based registry, the Haematological Malignancy Research Network (HMRN), to estimate outcomes for the comparator arm. People from the HMRN registry were chosen to align with the eligibility criteria of the zanubrutinib trials. Aggregate patient characteristics and anonymised time-to-event (progression-free and overall survival) individual patient-level data were obtained. The clinical experts considered that the data from the HMRN registry was representative of NHS clinical practice, except for the very small proportion of people who had fludarabine-based chemotherapy, which is not commonly used (see section 3.2). The committee concluded that the data collected from the HMRN registry was likely representative of standard care in the NHS.

The company presented a partitioned survival model with 3 health states: progression free, progressed disease and death. The probability of being in each health state was calculated using extrapolated progression-free and overall survival. People started the model in the progression-free health state. The model included a cycle length of 4 weeks with a half-cycle correction and a 27‑year time horizon. The EAG had concerns with using a partitioned survival model for relapsed or refractory MZL because of the independent modelling of progression-free and overall survival, and the relatively long time that people are in these health states. They noted that a state transition model may allow the condition to be modelled more accurately, but that data availability may limit the extent to which the model could be populated. The company confirmed that the data to populate a state transition model was not available. The committee concluded that the company's model structure was acceptable for decision making.

The company used patient-level data from MAGNOLIA‑003 to extrapolate progression-free and overall survival for zanubrutinib in the long term. The MAGNOLIA‑003 data was compared with the HMRN data via a MAIC (see section 3.6). For the comparator, progression-free and overall survival were extrapolated using the HMRN treatment data. The company used independently fitted survival models and selected the best-fitting distributions based on statistical fit, visual inspection and clinical plausibility. Because there was no evidence of a violation in the proportional hazards assumption between zanubrutinib and the HMRN treatments in progression-free and overall survival, the company considered it statistically appropriate to use the same distributions for both treatment arms. In the model, the company restricted overall survival by age- and gender-matched all-cause mortality for both treatment arms, such that the risk of death was never lower than in the general population. It also restricted progression-free survival by overall survival, such that people could not be progression free for longer than they were alive. The company's base case used the log-logistic distribution for progression-free and overall survival extrapolations for zanubrutinib and the HMRN treatments. The EAG had concerns about the immature survival data for zanubrutinib and the significant heterogeneity in extrapolations from different distributions for both treatment arms. The clinical experts considered that at 5 years, it was clinically plausible for 40% of people on the HMRN treatments to be alive. They considered that the log-logistic distributions provided the most plausible extrapolations. The committee noted that the latest data cut was in May 2022, and queried whether further data cuts were expected. The company reported that further data collection will only focus on safety and not efficacy outcomes. The committee agreed that the log-logistic distribution provided clinically plausible extrapolations and concluded that the company's base case was acceptable for decision making.

In the company's base case, it used the background mortality risk of the general population. It also provided a scenario analysis using an increased background mortality risk to reflect that people with relapsed or refractory MZL are likely to have an increased risk of death compared with the general population. The company used a standardised mortality ratio (SMR) of 1.41, previously applied in NICE's technology appraisal guidance on polatuzumab vedotin with rituximab and bendamustine for treating relapsed or refractory diffuse large B-cell lymphoma, to the background mortality in the model. The committee noted that the overall survival curves were restricted relatively early, with the risk of death replaced with the background mortality of the general population (see section 3.8). It considered that people who had relapsed or refractory MZL would likely have a greater risk of death than the general population. It concluded that the mortality risk of people with relapsed or refractory MZL would be higher than the age-matched general population and that an SMR of up to 1.41 should be used in the model.

In the company's model structure, the efficacy of zanubrutinib naturally waned over the time horizon with the hazard ratios of both progression-free and overall survival tending to 1. In its base case, the company assumed no additional treatment effect waning. At the request of the EAG at clarification, the company provided scenario analyses on additional treatment effect waning over varying periods. One of these was based on the median time to stopping zanubrutinib calculated in the model. The company considered it clinically inappropriate to assume that 50% of people would continue to have treatment without gaining any benefit from zanubrutinib. The clinical experts explained that relapses often occur continually in MZL and so there would be some treatment effect waning for both the zanubrutinib and HMRN treatment arms. They considered that zanubrutinib would likely have a longer time to relapse than the HMRN treatments, but agreed that there is limited evidence on the length of time it takes for a relapse to occur because of the immaturity of the zanubrutinib data. The committee considered that it was uncertain whether additional treatment effect waning should be modelled. It noted that there may be differential waning depending on whether people have zanubrutinib or the HMRN treatments. It concluded that the company's base case, which already accounted for some treatment effect waning, was appropriate for decision making.

In the EAG's base case, it accepted the company's comparator that included the HMRN treatments, the log-logistic distributions for long-term extrapolations of progression-free and overall survival (see section 3.8) and no additional treatment effect waning (see section 3.10). It used different utility values for the progressed disease health state based on NICE's technology appraisal guidance on lenalidomide with rituximab for previously treated follicular lymphoma, applied different disutility values and durations specific to each adverse event, and used eMIT prices. At the factual accuracy check, the company accepted the EAG's base case such that all assumptions were aligned in the company's revised base case.

The committee's preferred assumptions were largely in line with the company's revised model and the EAG's base case (see section 3.11). Except the committee considered that an increased background mortality risk should be applied to reflect that people with relapsed or refractory MZL are likely to have an increased risk of death compared with the general population (see section 3.9).

NICE's health technology evaluations manual notes that judgements about the acceptability of a technology as an effective use of NHS resources will take into account the degree of certainty around the incremental cost-effectiveness ratio (ICER). The committee will be more cautious about recommending a technology if it is less certain about the ICERs presented. The manual also states that decisions about the acceptability of the technology will consider aspects that relate to uncaptured benefits and non-health factors. The committee recalled the statements from the clinical and patient experts about the significant unmet need for effective and safe treatments in this rare condition. It also noted that zanubrutinib has a novel mechanism of action and, as an oral treatment, would be easily administered and fit into the existing care pathway. The committee acknowledged the high unmet need for novel treatments, but it also noted the high levels of uncertainty, including:

• the representativeness of the populations from MAGNOLIA and AU‑003, the immature progression-free and overall survival data and the lack of direct comparative trial evidence (see section 3.4)

• the limitations of the unanchored MAIC and lack of adjustment for all potential confounders and effect modifiers (see section 3.6)

• the uncertainty in the long-term extrapolations of progression-free and overall survival (see section 3.8).
  
  Balancing the unmet need and the uncertainties, the committee concluded that an acceptable ICER would be around the middle of the range NICE considers a cost-effective use of NHS resources (£20,000 to £30,000 per quality-adjusted life year [QALY] gained).

The committee considered the cost effectiveness of zanubrutinib compared with the HMRN treatments. In both the company's revised and the EAG's base case, and the scenario that included an increased background mortality risk (SMR of 1.41; see section 3.9), the deterministic and probabilistic ICERs were below the range the committee considered to be acceptable for this evaluation (see section 3.13). The exact ICERs cannot be reported here because some prices are commercial in confidence.

The committee did not identify any equality issues.

The committee considered if zanubrutinib was innovative. It did not identify additional benefits of zanubrutinib not captured in the economic modelling. So, the committee concluded that all additional benefits of zanubrutinib had already been taken into account.

NICE's advice about conditions with a high degree of severity did not apply.

The ICERs using the committee's preferred assumptions were below the range the committee considered to be acceptable for this evaluation (see sections 3.13 and 3.14). So, zanubrutinib is recommended, within its marketing authorisation, for treating MZL in adults who have had at least 1 anti‑CD20-based treatment.