Axonics sacral neuromodulation system for treating refractory overactive bladder

Two studies met the inclusion criteria defined by the scope:

• ARTISAN‑SNM (McCrery 2019 and Lane 2020; n=129 patients), which included 19 centres in the US and Europe (the Netherlands, Belgium, France and the UK) and had a follow up of 12 months.

• RELAX‑OAB (Blok 2018a and b, Blok 2019a and c; n=51 patients), which included 7 centres in Europe (the Netherlands, Belgium, France and the UK) and had a follow up of 2 years.
  
  Both studies were non-comparative, before and after, intra-patient, observational studies reporting patient outcomes as a change from baseline. Both studies had design and reporting weaknesses. The company also submitted evidence for the non-rechargeable comparator system.

ARTISAN‑SNM included people with urinary urge incontinence. Response to therapy was defined as at least a 50% reduction in urinary urge incontinence (episodes of urinary leaks) per day according to 3‑day urinary diary entries at 6 months. RELAX‑OAB included people with urinary urge incontinence (n=37) and all except one had symptoms of urinary frequency (n=50). Although not considered a primary outcome, response to therapy in RELAX‑OAB was defined as at least a 50% reduction in episodes of urinary leaks for people with urinary urge incontinence. For people with urinary frequency in RELAX‑OAB, response to therapy was defined as at least a 50% reduction in voids or reduction to fewer than 8 voids per day. ARTISAN‑SNM defined test response as symptoms responding to therapy at 1 month after implant. RELAX‑OAB defined test response as symptoms responding to therapy at 2 weeks or 1 month after implant.

ARTISAN‑SNM reported a statistically significant reduction in:

• mean daily urinary urge incontinence (episodes of urinary leaks) from 5.6±0.3 at baseline to 1.3±0.2 after 6 months, and 1.4±0.3 at 1 year (p<0.0001)

• urinary frequency episodes (average voids per day) from 11.6±0.3 at baseline to 8.7±0.2 at 6 months (p<0.0001).
  
  RELAX‑OAB reported a reduction in:

• mean daily urinary urge incontinence (episodes of urinary leaks) from 8.3±0.8 at baseline to 1.8±0.5 after 1 year (p<0.001) and to 1.7±0.5 at 2 years (p<0.0001)

• urinary frequency episodes (average voids per day) from 14.3±1.1 at baseline to 8.0±0.5 at 1 year (p<0.001) and 7.3±0.4 at 2 years (p<0.0001).
  
  The evidence suggests that symptoms of urinary urge incontinence were more likely to improve than urinary frequency symptoms, but this depended on how improvement was defined. The clinical effectiveness of Axonics sacral neuromodulation (SNM) system was not assessed beyond 2 years.

Both studies reported scores for the domains of the quality-of-life measure ICIQ‑OABqol before and after treatment. ARTISAN‑SNM reported an average score improvement of 34 points at 1 year and RELAX‑OAB reported an average improvement of 29 points at 2 years. Absolute before and after quality-of-life scores were not reported.

Because there was no long-term clinical evidence, a second independent external assessment centre (Newcastle EAC) assessed the company's technical data supporting the claimed 15‑year life span for the battery and the device's compatibility with MRI. The EAC concluded that, if moderate or typical stimulus is maintained for the lifetime of the battery, it is likely that the battery will exceed its claimed life span of 15 years. No evidence was submitted to explore the potential effect of lead migration (that is, when the lead moves from where it was inserted) or if the lead breaks. Minor lead migrations can be compensated for by increasing the stimulus, but major lead migration and breakage need to be corrected by surgery.

The de novo cost analysis used a Markov model adapted from a previously published model (Noblett 2017) to compare the rechargeable Axonics SNM system with the non-rechargeable system in people with overactive bladder. Most clinical parameters in Noblett 2017 are derived from the 12‑month Insite trial (Noblett 2016). The de novo model involved a quarterly progression between 3 health states (on SNM therapy, stopping therapy and death). The model assumed equivalent clinical effectiveness, therapy stopping rates and rate of adverse events for both the rechargeable and non-rechargeable systems, so the same inputs were used for both arms. Results showed that over a 15‑year time horizon, using Axonics SNM system in this population is cost saving by £6,038 per person.

The EAC considered that the study central to the company's model (Noblett 2017) had potentially serious limitations in reporting data sources. It noted that it was difficult to identify data stated in the model from the referenced papers. This limitation was partly addressed by an updated reference list submitted by the company. In the EAC's base case, clinical equivalence was not assumed so data on therapy stopping rate and lead migration were taken from Axonics SNM studies for the rechargeable SNM arm of the model. The company model assumed that all adverse events happen in the same cycle as the implant procedure. This was revised because lead migration, lead breakage and pain may not be related to the procedure and may continue throughout the life of the model. The assumption that the implant procedure would need an inpatient admission was changed to a day-case surgery, based on expert advice. A further revision was made to the cost model after consultation to apply the same rate of surgical site infection for both Axonics SNM system and the non-rechargeable comparator (1%).

In the EAC's preferred base case, Axonics SNM system was associated with an estimated cost saving of £6,025 per person over 15 years. A one-way sensitivity analysis showed that the timing of device replacement is a key driver of the cost savings. Threshold analysis showed that Axonics SNM system becomes cost saving when the life span of the technology is 6 years or longer.