Manufacturer’s Model Structure

For infantile-presentation patients, a survival model was used based on the prognosis in infantile-presentation patients as observed in the natural history study, LAL-1-NH01 (Jones et al. 2016),² with median age at death of 3.7 months (including those patients who received hematopoietic stem cell transplant [Figure 1]).

The model was structured as a Markov health state transition model. A lifetime horizon was used, and a monthly cycle length was chosen in order to reflect the timing of mortality (and associated neonatal intensive care) over the first 24 months of life. A half-cycle correction was applied in the first and last cycles of the model, to reflect transitions occurring throughout the cycle (i.e., as opposed to at the end of the cycle).

Additional health states were included in the model to account for intensive care medical resource utilization for patients who proceed to LAL deficiency-related death, and quality of life and background mortality in patients who do not proceed to LAL deficiency-related death. For neonatal intensive care, patients were assumed to spend a month in this state prior to LAL deficiency-related death; in sensitivity analysis, the length of time in neonatal intensive care was extended. The manufacturer assumed that patients surviving beyond 24 months take on general-population quality-of-life norms (based on UK values from Szende et al. [2014],⁷ as none are available for Canada) and survival (based on Canadian life tables from Statistics Canada [2017]).¹²

Figure 1Manufacturer Model Structure for Infantile-Presentation Patients

HS = health state; LAL = lysosomal acid lipase; QoL = quality of life.

Source: Adapted from manufacturer’s PE submission.³

Transition probabilities in the model for infantile-presentation patients include (i) the monthly mortality risk from LAL deficiency-related causes and (ii) the monthly mortality risk from background (i.e., non–LAL deficiency-related) causes. Monthly background mortality rates are modelled based on Canadian life tables available from Statistics Canada (2017).¹² The monthly mortality risk from LAL deficiency-related causes is based on survival observed in the natural history study LAL-1-NH01 and the VITAL study, as reported in Jones et al. (2017).⁴ Based on the observed mortality rates for infantile-presentation LAL deficiency patients aged 0 to 24 months, parametric survival modelling was also conducted to predict mortality risk for treated and untreated patients. Predicted survival was estimated under various assumptions regarding the survival distribution, including exponential (i.e., constant hazard function), Weibull and Gompertz (i.e., accelerated failure time with monotonic hazard functions), and log-normal (i.e., accelerated failure time with non-monotonic hazard function).³

For pediatric/adult-presentation patients, a model of liver disease progression was used. The manufacturer acknowledged that, while LAL deficiency threatens multiple vital organ systems including the liver, heart, spleen, and gastrointestinal tract, liver disease outcomes were focused on in the model, as they frequently occur early in the disease process, may have severe impact on quality of life and survival, and cannot be addressed with current supportive care.³ Therefore, the economic model focused on the hepatic aspect of LAL deficiency: the progression from hepatic fibrosis to compensated cirrhosis, decompensated cirrhosis, and hepatocellular carcinoma, with the potential for liver transplantation (Figure 2).

The manufacturer considered that non-alcoholic fatty liver disease, which involves deposition of fat (steatosis) in the liver due to causes other than excessive alcohol use, or its progressive form, non-alcoholic steatohepatitis (NASH), might be the best disease analogue for LAL deficiency.³ The manufacturer obtained confirmation from clinical experts who indicated that NASH is the most appropriate disease analogue, and that disease progression up to the point of cirrhosis would likely differ in LAL deficiency versus NASH, but from the point of cirrhosis onwards, it would likely be similar.

As such, in the cost-effectiveness analysis of sebelipase alfa for LAL deficiency, the model structure for advanced liver disease health states was derived from the consensus in NASH cost-utility studies.³ In addition, the fibrosis state included in such studies is expanded to reflect the METAVIR fibrosis stages (i.e., F0 to F4) to capture the impact of treatment on timing of progression to advanced liver disease states.

The model was structured as a Markov health state transition model. A lifetime horizon was used, and an annual cycle length was chosen, consistent with other models of NASH. A half-cycle correction was applied in the first and last cycles of the model, to reflect transitions occurring throughout the cycle (i.e., as opposed to at the end of the cycle). Background mortality rates were modelled based on Canadian life tables available from Statistics Canada (2017).¹² Age at baseline was assumed to be 11 years of age and was based on (i) the mean age of onset reported in Bernstein et al. (2013)⁸ of 5 years, plus (ii) the time between mean age of onset and age of diagnosis reported in Burton et al. (2015) of 6 years.⁵

Figure 2Manufacturer’s Model Structure for Pediatric/Adult Presentation

F0 to F4 = METAVIR fibrosis stages; DCC = decompensated cirrhosis; HCC = hepatocellular carcinoma.

Note: Transitions possible due to the sebelipase alfa treatment effect (i.e., regression between fibrosis stages) are reflected by blue, dashed arrows.

Source: Adapted from manufacturer’s PE submission.³

The manufacturer calculated the fibrosis progression rate (FPR) as reflecting the average number of fibrosis stages progressed per year, and used the FPR to model patient transitions between stages on the F0 to F4 range. The FPR for pediatric/adult presentation LAL deficiency was calculated based on biopsy data from the ARISE trial, Bernstein et al. (2013),⁸ and Burton et al. (2015).⁵ In order to parametrize transitions to advanced liver disease, estimates of transition probabilities from Crossan et al. (2015)⁹ and a systematic literature review performed by the UK’s National Health Service National Institute for Health Research, were considered.³

Manufacturer’s Results

The manufacturer’s deterministic base-case results are presented in Table 13.

CADTH Common Drug Review Reanalyses

Given the limited clinical information available on patients with LAL deficiency and the uncertainty associated with the clinical evidence on the efficacy of sebelipase alfa for infantile and pediatric/adult presentations of LAL deficiency, CDR undertook exploratory analyses that tested alternate assumptions on the efficacy of sebelipase alfa on halting or improving liver disease in pediatric/adult-presentation patients with LAL deficiency as well as alternate assumptions on patient weight changes over time, model time horizon, and health state utility based on a search of the available literature and based on feedback from the clinical expert.

CDR conducted an exploratory analysis that varied the patient weight at 21 years of age and beyond applied an annual 1% increment to the patient’s weight from the age of 21 to 60, and an annual 1% decrement in weight thereafter. The results of the exploratory analyses are presented in Table 14 and show that the total costs with sebelipase alfa had increased based on the increasing total drug costs as the patient gains additional weight up to the age of 60 years.

Due to model limitations, CDR was unable to vary the duration of treatment effects in the submitted models; however, conducted exploratory analyses that varied the model time horizon between 1 and 20 years for both patient presentations. The results of these analyses are in Table 15 below.

Based on the uncertainty with the key efficacy outcomes in the ARISE trial being surrogate outcomes instead of hard clinical outcomes, and based on the FDA medical review’s indication that “normal ALT does not necessarily exclude the presence or progression of liver disease,”⁶ and that “normalization of ALT does not reliably represent a clinical benefit,”⁶ exploratory analyses were conducted by CDR. In these exploratory analyses, CDR modelled efficacy scenarios for sebelipase alfa in pediatric/adult-presentation patients with LAL deficiency in which liver disease progression is halted but may then either remain stable, an optimistic scenario based on available clinical evidence, or may progress by 10% each cycle. The results in Table 16 and Table 17 show that ICUR values for sebelipase alfa in pediatric/adult-presentation patients are not significantly impact by the assumption of disease stability or progression.

An exploratory analysis was conducted by CDR that used utility values from a published economic evaluation in patients with non-alcoholic steatohepatitis.¹⁰ The results of the CDR exploratory analysis did not significantly impact the results (Table 18).

The CDR exploratory multi-way analysis consisted of varying the patient weight by 1% in the infantile and pediatric/adult-presentation LDL and using the optimistic scenario of sebelipase alfa effectively stabilizing liver disease progression in pediatric/adult-presentation patients (Table 19).