1
Thesis Master Business Administration Health
The economic evaluation of the orphan medicine
177Lu-DOTATATE
for patients with an advanced Midgut Neuroendocrine Tumour
in the Netherlands.
A simulation Study
Student: M.E. Hagendijk (Marije) Student number: s3473546
Supervisors & Co-supervisor: Maarten Postma, Simon van der Pol & Jochen Mierau Date of submission: 24 June 2019
Word count: 6983
Abstract
Background – The 177Lu-DOTATATE treatment is a type of peptide receptor radionuclide therapy treating
Neuroendocrine Tumours (NETs). These NETs have a very low incidence. That is why the 177Lu-DOTATATE
treatment gained the status ‘orphan medicine’ in 2018. This status gives the pharmaceutical company, Advanced Accelerator Applications a Novartis Pharmaceuticals company (Novartis International AG, Basel, Switzerland), a monopoly position for selling the treatment for ten years within the European Union. After receiving this patent, the pharmaceutical company doubled the selling price of the treatment by six.
Objectives – Because of this increase in price, multiple stakeholders of the Dutch healthcare system are
worried the 177Lu-DOTATATE treatment will no longer be reimbursed by the Dutch healthcare insurers. In this
study an economic evaluation is executed to capture the impact of the increase in price on the cost-effectiveness of the 177Lu-DOTATATE treatment.
Method – Data used in this evaluation is distracted from the NETTER-1 study, a randomized control trial,
comparing two groups: a group receiving 177Lu-DOTATATE injections plus best usual care, Octreotide LAR, and
a group receiving a high dose of the best usual care. A Markov Model is developed to evaluate the costs per life-years gained (LYG) and per quality-adjusted life-years (QALY). A healthcare perspective is used. This model included three health states: the stable disease, progressive disease and death. A scenario analysis is
conducted to test the differences in cost-effectiveness between the old price and the base-case.
Results – In the base-case analysis the cost-utility analysis gave a deterministic incremental cost-effectiveness
ratio (ICER) of €55.696,62/LYG and €72.176,85/QALY, while the old treatment price had a deterministic ICER value of €17.370,92/LYG and €22.510,85/QALY. The sensitivity analysis of this value showed that for the new price there was a 70,2% probability of cost-effectiveness while the old price had a 100% certainty, assuming the Dutch willingness-to-pay (WTP) threshold of €80.000,-/QALY.
Conclusions – The new price of the 177Lu-DOTATATE injections have a much lower probability to be
cost-effective at the WTP threshold. The concerns of the stakeholders, about the reimbursement of this treatment in the Dutch healthcare system, are justified.
2
Introduction
Neuroendocrine tumours (NETs) are slow-growing tumours located in the hormone-producing cells of the body, as the Pancreas and the Midgut. This type of tumour is very rare, between 1990 and 2010 the recorded incidence was only 47.800 patients within the Netherlands (Korse, Taal, van Velthuysen, & Visser, 2013).
Since the 90s new types of treatments for NETs, peptide receptor radionuclide therapy (PRRT), are developed. This kind of treatment target the somatostatin receptors (SSR) on the surface of the hormone-producing cells, connect with these SSRs and its radioactive character provides local radiotherapy to the NET sites. At this moment, the only PRRT allowed on the European pharmaceutical market is the 177Lu-DOTATATE treatment, a combination of Lutetium-177 (IDB Holland BV, Baarle-Naussau, Netherlands) with the amino acid peptide DOTA-octreotate. This medicine, also called Lutetium-Octreotate, is developed by researchers since 1985 and used for treatment since 1992 in the Erasmus Hospital in Rotterdam, the Netherlands (Biosynthema Inc.). In 2011 a small French pharmaceutical company Advanced Accelerator Applications (AAA) became the owner of 177Lu-DOTATATE. They financed the research and development costs for a randomized control trial (RCT) phase III executed between 2012 and 2015, published in 2017 as the NETTER-1 study (Strosberg et al., 2017). Because of the earlier publications of case reports by Biosynthema Inc (Brabander et al., 2017; Kwekkeboom et al., 2008; Kwekkeboom et al., 2005), the additional phase III RTC provided enough evidence for the European Medicines Agency to give 177Lu-DOTATATE market authorisation. In the same year, only a few days after AAA was bought by a big Swiss pharmaceutical company, Novartis Pharmaceuticals (Novartis International AG, Basel, Switzerland), the 177
Lu-DOTATATE treatment obtained the status of being an orphan medicine. This status, ‘orphan medicine’, is assigned to medicine treating rare conditions in small patient populations. This status brings new privileges for the pharmaceutical company, to give them the opportunity to recoup the research and development costs even with a small target group.
Novartis received a patent for ten years in the European Union and seven years in the United States of America. Meaning that this pharmaceutical company has a monopoly position for selling this orphan medicine these amounts of years. That is why, in January 2019, Novartis was able to raise the 177Lu-DOTATATE treatment price from €16.000,- to €90.000,- (Hordijk, 2019).
The Dutch national cancer organisation and Dutch healthcare insurers accuse Novartis of making abuse of their monopoly position. They assume that because of this increase in price, the medicine will no longer be reimbursed for the patients in need by the Dutch healthcare insurers. In January 2019 the Dutch Minister for Medical Care stated during an interview: ‘I told Novartis that I am not willing to pay their price’. Novartis claims that the medicine is proved to be cost-effective in other European countries with the new price(NU.nl/ANP, 2019), no publications were found.
3
Method
The evaluation
The purpose of this article is to evaluate the cost-effectiveness of 177Lu-DOTATATE treatment compared to a treatment with a high dose of best supportive care (BSC), Octreotide LAR. The 177Lu-DOTATATE is a treatment consisting of four injections, once every 8-10 weeks. Besides these four injections, the patients receive also octreotide LAR 30mg injections 24 hours after the 177 Lu-DOTATATE injection and monthly after the completion of all four infusions. These Octreotide LAR injections are used for symptom control. The control group received injections of Octreotide LAR 60 mg every ±4 weeks (28 days). Both treatments, including the BSC, stopped when the patients showed progressive disease measured by a CT scan or MRI.
Patient population
The data included in this economic evaluation was derived from the patient population studied in the NETTER-1 study (Strosberg et al., 2017). This patient population in the NETTER-1 study (N=229) was randomly divided in the two study groups: 177Lu-DOTATATE treatment plus BSC (N=116; 54% male) and the control group (N=113; 47% male). The 177Lu-DOTATATE group had an average age of 63 (±9) years and the control group 64 (±10) years. The most common primary tumour site in both groups was the Ileum (74%;73%) and the most common site of metastasis in both groups was the liver (84%;83%).
Model design and Time Horizon
For this economic evaluation a simplified Markov model with three health states was designed (Figure 1); stable disease (SD), progressive disease (PD) and the absorbing death state. The SD health state contains that there is no present growth in tumour size of even a reduction in the tumour size. When growth in tumour size is visible by a CT scan or MRI, the PD health state is reached.
According to the CTCAEv3.0 grade III and IV adverse events (AE) can be severe and life-threatening, the influence of these AE on the costs and health outcomes were included in the SD health state (Trotti et al., 2003).
Within this Markov model a cohort of 1.000 patients suffering an Advanced Midgut Neuroendocrine Tumours was modelled. This model had a time horizon which was modelled lifelong (≥99% in absorbing state), each cycle time contains of 28 days. For each cycle the quality-adjusted life years (QALY), life years gained (LYG) and total costs were given. With this information the cost-utility analysis and a LYG-analysis were conducted. The difference in total costs divided by the difference in total QALY or LYG between the two cohorts receiving the 177Lu-DOTATATE treatment or high-dose octreotide LAR gave the incremental cost-effectiveness ratio (ICER). This Markov model is designed and build within Microsoft Excel® 2016 (Microsoft Corp, Redmond, WA).
Transition probabilities
4
these data-points from the KM-curves. With the help of Microsoft Excel® 2016 (Microsoft Corp, Redmond, WA) an exponential function was found for PFS and a linear function for the OS. A hazard ratio (HR) including the 95% confidence interval (95% CI) is given for the PFS of the 177 Lu-DOTATATE group compared to the control group, for the OS a HR is given without the 95%CI in the NETTER-1 and the first update, see table 1 (Strosberg et al., 2017; J. R. Strosberg et al., 2018). For the PFS the exponential function of the control group was used to calculate the probability per cycle time of 28 days, because of the exponential character the probability was assumed to be stable over time. The given HR (95% CI) was applied to this probability to determine the probability
(mean(±95%CI)) for each cycle time of the 177Lu-DOTATATE group (table 2).
In the article of Arnold and his colleagues (2005) a HR (±95% CI) is given for the OS in the health state SD compared to the PD (table 1), this knowledge is used to create two different OS
probabilities for the control group; the overall survival for the PD and the overall survival for the SD health state. The mean in probability of these two curves is equal to the probability of the linear curve of the original OS KM-curve. Then again, with the HR found for the 177Lu-DOTATATE group compared to the control group in the first update of the NETTER-1 study (J. R. Strosberg et al., 2018), the probabilities were set towards the absorbing state (table 2).
Because of the older age of the patient population (mean=63.5 years), the probability to die due to other causes than the NET is high and increases over time. The age dependent survival probabilities for the overall population were distracted from the Dutch Statistics of public health
(Volksgezondheid en Zorg info, 2017). For all OS probabilities of both groups in the Markov model was assumed that when the age dependent probability to die was higher than the probability to die according to the follow up in the NETTER-1 study, the age dependent survival probability was assumed in the model as the probability to die.
To take the uncertainty around the probabilities into account, a beta distribution is included for the sensitivity analysis. A beta distribution samples within the interval 0-1, based on the α and β
parameters (Briggs, 2001). Besides this, for the parameters with an unknown upper and lower 95%CI an 80%-120% interval of the deterministic value is included.
5
Table 1. Effects (HR) used for the probabilities.
HR (±95% CI) Source
PFS * 0.21 (0.14-0.33) (Strosberg et al., 2017)
OS* 0.41 (not given) (J. R. Strosberg et al., 2018)
OS SD vs. OS PD** 0.44 (0.24-0.79) (Arnold et al., 2005)
* Effect of 177Lu-DOTATATE treatment compared to the high dose of BSC treatment.
**Effect of BSC treatment OS stable disease compared to BSC treatment OS progressive disease.
CI = confidence interval. OS = overall survival. HR = hazard ratio. PFS = progressive free survival. SD = stable disease. PD = progressive disease.
Table 2. Transition probabilities of the first cycle (mean age = 63,4y) assumed in the Markov model.
Transition Patient group Probability (±95% CI)~
SD – PD** High BSC dose .0854
177Lu-DOTATATE + BSC .0183 (.0119 - .0283)
SD – DEATH* High BSC dose .0095
177Lu-DOTATATE + BSC .0073 (.0059 - .0088)+
PD – DEATH* High BSC dose .0216
177Lu-DOTATATE + BSC .0116 (.0093 - .0139)+
*Exponential probability over time **Constant probability over time +Assumed 80%-120% interval
~Beta distribution.
Utilities
The utilities for the different health states are distracted from the Time Trade-Off study of Swinburn and colleagues (2012). In this study patients with a NET tumour filled in the EQ-5D, members of the general public performed the health state valuation of these health states given by the patients (Swinburn, Wang, Chandiwana, Mansoor, & Lloyd, 2012). The outcomes of this study provided the utilities for: stable disease without AE, stable disease combined with one kind of grade III or IV AE and progressive disease. So, in both treatment groups the same utilities for SD without AE and PD are assumed. The utilities of the SD with AE for the two treatment groups were calculated based on the percentages experiencing the different grade III and IV AE (Strosberg et al., 2017). Note that Swinburn et al. (2011) did not present utility values of all occurring grade III and IV AE of the treatments mentioned in the NETTER-1 study. The utility value included in this model is only based on the AEs of which the utility values could be distracted from Swinburn et al (2012). The utilities for the different health states of the Markov model can be found in table 3.
In the 177Lu-DOTATATE treatment group 41% of the patients and in the high dose BSC group 33% of the patients experienced a grade III of IV AE during the trial. In the paper, its supplementary appendix and the first update lacking information is given about the timing of the occurrence of these adverse events (Strosberg et al., 2017). That is why, in this economic evaluation, the
6
Table 3. Utility values per health state (Swinburn et al., 2012)
AE = adverse events. SD = stable disease. PD = progressive disease. BSC = best supportive care. Costs
This evaluation is executed from the healthcare perspective, no indirect costs as productivity losses or volunteer time were included because of lacking data of these contents from patients with advanced Midgut Neuroendocrine Tumours. Medicine costs, costs to manage adverse events, medical resource utilization costs and indirect medical costs as end-of-life costs were considered. The current medicine costs at the Dutch pharmaceutical market included in this evaluation, except the 177Lu-DOTATATE treatment, are derived from the website of the Dutch healthcare institute (Zorginstituut, Medicijnkosten.nl). The costs of the 177Lu-DOTATATE treatment were derived from the Dutch magazine of medicine (Hordijk, 2019). During the 177Lu-DOTATATE treatment the patients receive before every injection a CT scan or MRI.
When patients shift to the PD health state the 177Lu-DOTATATE and Octreotide LAR treatment costs reduce to zero, according to the NETTER-1 study. The treatments stop at this point, the follow-up appointments are the cause of the remaining health care costs. These outpatient clinic
appointments have a frequency of every 6 months, where blood testing is executed (table 4). During this follow-up period no other data of cost sources are known.
The medical resource costs are derived from the Dutch healthcare cost manual (Hakkaart-van Roijen, Van der Linden, Bouwmans, Kanters, & Tan, 2015). These cost-values given in this manual were from 2014, because of the changing cost-values over the years the cost-values from 2014 were multiplied once by the price index from 2014 to 2019 of 1,067 (Statline, CBS). Age-dependent indirect medical costs are derived from the Dutch PAID 1.1 tool (van Baal et al., 2011). The costs of Pancreas cancer was selected as the direct medical costs, all other ‘indirect medical’ costs per other-cycle time and last-cycle time (end-of-life) were measured.
Because of the low toxic effects and the transient character of most of the AE during the 177 Lu-DOTATATE treatment the extra costs for managing the AE were low. These extra costs were based on the data of the Safety Assessments from the Supplementary Appendix of NETTER-1 (table 4). For all costs included in this model the assumption was made that the standard error of the costs is equal to the mean sampled by a Gamma distribution (Briggs, Sculpher, & Claxton, 2006). This assumption was used because there is a lack of information about the uncertainty within the costs.
Health state Mean utility value Min-Max Distribution
SD without or grade I/II AE 0.771 (0.509-0.818) Beta
SD with grade III/IV AE
Octreotide LAR 60 mg 0.666 (0.619-0.709) Beta
177Lu-DOTATATE + BSC 0.687 (0.642-0.730) Beta
PD 0.612 (0.564-0.659) Beta
7
Table 4. Overview costs
Actions per group
Costs per cycle per capita (€)
When?
Source
177LU treatment + BSC Medicine costs
Amino Acids Infusion+ Granisteron 3mg+ Hotel costs (per night)+ MRI+ CT+ AE control+ Injection BSC 30 mg+ GP visit+ 23.000,- (Base-case) 4.000,- (Old-price scenario) 89,38 62,13 685,78 473,82 304,14 9386,78 537,25 35,22 SD cycles T=0;2;4;6 SD cycles T=0;2;4;6 SD cycles T=0;2;4;6 SD cycles T=0;2;4;6 SD cycles T=0;2;4;6 SD cycles T=12+6 T=0 SD cycles T=0;2;4;6 +1 SD cycles T=0;2;4;6 +1 (Hordijk, L., 2019) Medicijnkosten.nl Medicijnkosten.nl
(Hakkaart-van Roijen et al., 2015) (Hakkaart-van Roijen et al., 2015) (Hakkaart-van Roijen et al., 2015) (Strosberg et al., 2017)
Medicijnkosten.nl
(Hakkaart-van Roijen et al., 2015)
Octreotide LAR (BSC) Injection BSC 60 mg+ GP visit+
CT scan+
Visit outpatient clinic+
1074,50 35,22 304,14 180,85 SD every cycle SD every cycle - T=0;3;6;9;12 +6 SD cycles T=0;3;6;9;12 +6 SD cycles T=0;3;6;9;12 +6 Medicijnkosten.nl
(Hakkaart-van Roijen et al., 2015) (Hakkaart-van Roijen et al., 2015) (Hakkaart-van Roijen et al., 2015)
Follow up both groups Visit outpatient clinic+ Blood tests+
180,85 11,77
PD cycles T=0 +6 PD cycles T=0 +6
(Hakkaart-van Roijen et al., 2015) (Hakkaart-van Roijen et al., 2015)
Indirect medical costs
Indirect medical costs (other-month)+ End-of-Life (last-month)+
431,64* 46.720,56*
SD+PD every cycle
Death cycle N – Death cycle N-1
(van Baal et al., 2011) (van Baal et al., 2011)
8 Discounting
Discount rates for costs of 4% (0.33% per month) and for health of 1.5% (0.12% per month) per year are taken into account according to the Dutch pharmacoeconomic guidelines (Tan et al., 2012). A sensitivity analysis was performed with the same discount rate for both, costs and health. The upper value is an equal discount rate of 4% per year, the lower value is an equal discount rate of 0% per year.
Willingness to pay
The deterministic ICER gives the ratios difference in costs and the difference in QALY or LYG gain between the two treatment groups. A positive ICER can have two meanings: more costs but also more QALY/LYG or less costs but also less QALY/LYG for the studied intervention. In these two cases it is not directly clear which intervention to choose, it depends on the willingness to pay (WTP) of the healthcare system which treatment will be reimbursed.
The Dutch healthcare system makes use of different cost-effectiveness thresholds. These different thresholds depend on the burden of disease the intervention is treating. Cancer treatments have the highest WTP threshold because of the high burden of disease. This threshold gives that the Dutch healthcare system can afford €80.000,- per QALY or LYG unit (Zwaap, Knies, Van der Meijden, Staal, & Van der Heiden, 2015).
Sensitivity analysis
Insight in the influence of the uncertainty of the parameters included in the calculation of the deterministic ICER is given by the execution of a probabilistic sensitivity analysis (PSA); within Microsoft Excel® 2016 this is done by the execution of a Monte Carlo simulation. This simulation samples 1.000 time values from the distributions of the different input parameters giving different ICER values. Within this model there are two base-case PSA executed, one for the cost-utility analysis and one for the LYG-analysis.
Because of the focus in this study on the increase in costs of the 177Lu-DOTATATE injections, a scenario analysis is performed for the old and the new price. In this scenario analysis the PSA of the base case, new price, is compared with the PSA outcome when including the old price of the treatment.
A second scenario analysis is executed for the base-case scenario of the cost-utility analysis when excluding the uncertainty in costs. This scenario analysis give insight in the effects of the assumption made for the uncertainty in costs (Briggs et al., 2006).
The ICER values as outcome of the Monte Carlo simulation for the base-case and the old price are plotted in the CE-plane. All outcomes of the executed PSA were translated into cost-effectiveness acceptability curves (CEACs). These CEACs give insight in the probability that a treatment is cost-effective at a certain WTP threshold (Fenwick, Claxton, & Sculpher, 2001) .
9
Results
Deterministic analysis
See table 5 for the results of the deterministic cost-utility analysis and the life-years gained analysis of the base-case. The ICER of the cost-utility analysis is €72.176,85/QALY, this ICER has a positive NMB (€10.994,63) which means that it is located in the north-east quadrant of the CE-plane. Less costs are concerned with saving life-years, €55.696,62/LYG (NMB=€44.262,36).
Table 5. Results of deterministic cost-utility and life-years-gained analysis (discounted).
Variable High dose Octreotide LAR 177Lu-DOTATATE + BCS Difference ICER
Costs (€) 71.654,61 173.091,71 101.437,09
LYG* 3,55 5,37 1,82 €55.696,62/LYG
QALY* 2,29 3,70 1,41 €72.176,85/QALY
*Total from the start of treatment until end of lifelong (>99% in absorbing state), mean age of both groups was 63,5y. ICER = incremental cost-effectiveness ratio. QALY = quality adjusted life-years. LYG = Life-year gained.
CE-plane and CEAC cost-utility analysis
In figure 2 the distribution of the outcome of the Monte Carlo simulations is presented for the scenario analysis of cost-utility analyses of the case compared to the old price. For the base-case scenario an ICER is found of €72.176,85/QALY (95%CI=€37.087,99; €104.447,40) and
€55.696,62/LYG, for the old price scenario an ICER is found of €22.510,85/QALY (95%CI= -€11.468,-; €55.065,81) and €17.370,92/LYG.
Figure 2. Cost-effectiveness plane. Including the outcome of the Monte Carlo simulation of the current price of
177Lu-DOTATATE injections (light green), and the Monte Carlo simulations of the old price of the 177
Lu-DOTATATE injections (dark green). The deterministic ICERs for both are represented by a red dot. The black line represents the willingness-to-pay of €80.000,-/QALY.
10
Figure 3. Cost-effectiveness acceptability curve (CEAC) representing the probability of the cost-effectiveness at
a certain willingness to pay (WTP). LYG = life-years gained.
Figure 3 gives the CEACs of the base-case cost-utility analysis, the old price scenario cost-utility analysis, the scenario analysis where the cost uncertainty is not included and the base-case LYG analysis. At the threshold of €80.000,-/QALY the probability that the base case is cost-effective is 70,2%, while the probability of being cost-effectiveness at the old price for this threshold is 100%. The CEAC which excludes cost uncertainty shows a steeper slope, the probability being cost-effective in this case is 99,8% at the WTP threshold. The probability that the LYG-analysis is cost-effective at the threshold is 98,7%.
Figure 4. Tornado diagram of the deterministic sensitivity analysis, showing the effects of the upper and lower
values of the input parameters on the ICER outcome (base case=€72.176,85). - - - =WTP threshold. SD = stable health state. PD = progressive health state. AE = adverse events grade III/IV. HR = hazard ratio.
61000 66000 71000 76000 81000 86000 91000 96000 101000 Utility SD HR OS PR (80%-120%) Discount rate (0%-4%) HR OS SD (80%-120%) HR PFS Utility PR Utility 177Lu SD-AE Utility Control SD-AE
ICER [€/QALY] Low value High value 0 10 20 30 40 50 60 70 80 90 100 0 20000 40000 60000 80000 100000 120000 140000 160000 Pro b ab ili ty cos t-ef fe ctiv en e ss [ % ] WTP [€] price=23000 price=4000 LYG analysis
11 Deterministic sensitivity analysis
The outcomes of the deterministic sensitivity analysis is shown in figure 4, the Tornado diagram. The uncertainty interval of the utility value of the SD health state has the largest influence on the
deterministic ICER (€109.496,-; €68.018,17). The deterministic ICER value is in a smaller extent sensitive for changes in the other utility parameters. Besides the sensitivity for the SD utility, also the assumed uncertainty in the HR of the OS are large influencers (PR €61.788,66; €83.140,02 :ST
€64.095,44; €81.658,66). An equal discount rate of 4% gives an ICER of €82.243,39 and no discounting gives an ICER of €71.692,32.
Discussion
In this economic evaluation a cost-effectiveness analysis is executed for the 177Lu-DOTATATE treatment compared to a high dose Octreotide LAR intervention for patients with an advanced Midgut NET in The Netherlands. The focus was mainly on the big increase in price per 177 Lu-DOTATATE injection over the past year, implemented by the pharmaceutical company. The results show that the costs per QALY gained were €72.176,85 and per LYG €55.696,62, compared to the Octreotide LAR intervention. These values are located in the north-east quadrant of the CE-plane (figure 2), implying that the base-case of the 177Lu-DOTATATE treatment is more effective but is also more costly compared to the control group. The same counts for the scenario analysis of the old price, the costs per QALY gained were €22.510,84 and per LYG €17.370,92. This scenario analysis shows a difference in price of almost €50.000,- per QALY and €38.325,- per LYG which needs to be paid extra by the health insurers when applying the price level of the base-case.
The positive effect on health by the 177Lu-DOTATATE treatment is also supported by the findings in earlier studies including various NET patient sub-groups (Bodei et al., 2011; Brabander et al., 2017; Esser et al., 2006; Ezziddin et al., 2014; Kwekkeboom et al., 2008; Kwekkeboom et al., 2005). However, to our knowledge this is the first economic evaluation executed for the 177Lu-DOTATATE treatment for the Dutch healthcare system. Novartis claimed that in other European countries the 177Lu-DOTATATE treatment is proved to be cost-effective with the current price, but no publications are found to support this (NU.nl/ANP, 2019). Only one publication of a cost-effectiveness analysis comparing the NET treatments Everolimus, Sunitinib and 177Lu-DOTATATE is found (Mujica-Mota et al., 2018).
The PSA showed that the cost-utility base-case analysis had a cost-effectiveness probability of 70,2%, and 98,7% for the LYG-analysis at a WTP threshold of €80.000 per unit. The cost-effectiveness probability of the old price scenario was in both analyses 100% certain at this threshold. The substantial reduction in cost-effectiveness probability for the cost-effectiveness analysis (-29,8%) makes the consideration for reimbursement uncertain. The outcomes of a budget impact analysis in this case can help decision makers to make a final decision.
12
Besides this assumed uncertainty in the included costs, there were also some related cost items not included at all because of lack of information, as productivity losses and over the counter costs made by the patient itself (Sculpher & Drummond, 2001). Because of this only a healthcare perspective could be executed, while the Dutch guidelines recommend the social perspective (Byford & Raftery, 1998; Tan et al., 2012). Researchers executing medical trials need to become more aware of the importance of this information, and questionnaires about these indirect costs need to be
implemented more in the trials. In this way, the possibility will be created that the social perspective can be implemented in future research.
Another limitation in this study related to the costs, is the low amount of included costs during the PD health state (table 4). Besides the standard follow-up outpatient clinic visits, no other data of cost sources are known. Because this lack occurs in both the control and the 177Lu-DOTATATE group the impact is not expected to be very large.
The assumption that all adverse events occur at T=0 also caused a less precise evaluation, because of this assumption no discounting for the AE over time is included. This influenced the outcome of the total amount of discounted QALYs (table 5). Health-related quality of life of the patient population from the NETTER-1 study is published (J. Strosberg et al., 2018). Nevertheless, the utility values distracted from Swinburn and his colleagues (2012) were preferred, because of the clear distinction between the health states in this model. During calculating the overall utility value for the SD with grade III or IV AE per intervention group, only the AE are included of which the utility values were known (Casciano et al., 2012; Swinburn et al., 2012). Utility values of some AE were unknown, for example the AE abdominal pain, and so is not included in the calculated AE ratio. This could lead to an over- or underestimation of the utility value for SD with AE. The deterministic sensitivity analysis (figure 4) showed that the model is robust for chances in the utility values for SD with AE, so the influence of above on the outcomes are expected to be low.
In this economic evaluation a large variety in patient characteristics is included by the wide patient population of the NETTER-1 study. For future research it would be interesting to execute a sub-group analysis by patient characteristics as gender and age, the primary tumour site or the site of
metastasis (Strosberg et al., 2017). A limitation in this regard, by the NETTER-1 study as data source for this economic evaluation, is that the control group in clinical trials of alternative treatments are not comparable to the high dose Octreotide LAR control group. Clinical trials published, for example trials for Everolimus and Sunitinib treating pancreatic NETs, have a control group receiving best supportive care (Raymond et al., 2011; Yao et al., 2011). In 2018 there is executed a cost-effectiveness analysis by Mujico-Mota and his colleagues for the treatments Everolimus, 177 Lu-DOTATATE and Sunitinib. This comparison raises questions given the large amount of confidential information and the differences in characteristics of the control groups (Mujica-Mota et al., 2018). For this, it is desirable to conduct further research with comparable patient groups.
The Dutch guidelines suggest making use of the non-equal discount rate for health and costs (Brouwer, Niessen, Postma, & Rutten, 2005; Tan et al., 2012). But a lot of other countries still make use of an equal discount rate (Gravelle & Smith, 2001; NICE, 2008). When applying an equal discount rate of 4% to this model, the deterministic ICER shifts to €82.243,39, which is above the WTP
13
The datapoints within the KM-curves from the NETTER-1 study had a follow-up period of 18 months (Strosberg et al., 2017). The use of the manual Web Plot Digitizer to distract these points from the KM-curve can be the source of a small unregulated uncertainty. Besides this, the exponential and linear functions based on these data-points and their probabilities are only a reproduction of the real-world situation. In addition, the use of the HR from Arnold et al. (2005) to make a distinction between the death rate of the patients within the stable disease and the progressive disease health state, strengthens this reproduction of the real world. During calculating these probabilities from the two transient states towards the absorbing state, the mean of these two is assumed to be the original data. This assumption is made, while it is known that the distribution of patients is not equal in the two health states at all times. This could be the cause of deviations compared to real world data. The HR comparing the two 177Lu-DOTATATE groups with the Octreotide LAR controls, the influence on the final deterministic ICER value are not expected to be very large.
This reproduction of the real world is not different in other economic evaluations, a (Markov) model is always a simplification of the real-world situation. In a Markov model a time aspect per patient is not included, it represents a random process evolving over time (Briggs & Sculpher, 1998). In this study only three health states are included, while in real world more can be distinguished. For example, stable disease could be subdivided into complete response, partial response and stable disease. This model is simplified because no utility values were known for these health states separate. Future research can investigate these unknown values and their influence on the outcomes of the economic evaluation.
After reviewing the model and the outcomes of this economic evaluation, one question remains: Were the Dutch national cancer organisation and Dutch healthcare insurers right when accusing Novartis of making abuse of their monopoly position when increasing the treatment price? The monopoly position of ten years, by obtaining the status of orphan medicine, is granted to give pharmaceutical companies the opportunity to earn back the research and development (R&D) investments. Already shown in literature is that this monopoly position for pharmaceutical companies is very tempting to drive costs to the maximum, even when this is not necessary to compensate the R&D costs (Sheldon, 2012). In this case the pharmaceutical company makes abuse of the given situation. A price increase in the extent of the 177Lu-DOTATATE injections raises
expectations of such abuse. In future research a follow-the-money analysis needs to be conducted to determine this accusation.
As long as there is no concrete decision about reimbursement or not for the 177Lu-DOTATATE
14
Conclusion
The deterministic costs per QALY gained were €72.176,85 and per LYG were €55.696,62 for the new-priced 177Lu-DOTATATE injections, compared to the Octreotide LAR intervention. These deterministic values are below the willingness-to-pay of €80.000,-. Because of uncertainty in this model, the probability of having a probabilistic ICER value below the €80.000,- is 70,2% for the QALY’s and 98,7% for the LYG. While, with the old price of the 177Lu-DOTATATE injections these deterministic costs per QALY and LYG were much lower and the probability of being cost-effective for both was 100%. Because of the reduction in the cost-effectiveness probability the decision of reimbursement will depend on the budget impact analysis.
References
Arnold, R., Rinke, A., Klose, K.-J., Müller, H.-H., Wied, M., Zamzow, K., Moll, R. (2005). Octreotide versus octreotide plus interferon-alpha in endocrine gastroenteropancreatic tumors: a randomized trial. Clinical Gastroenterology and Hepatology, 3(8), 761-771.
Bodei, L., Cremonesi, M., Grana, C. M., Fazio, N., Iodice, S., Baio, S. M., Sansovini, M. (2011). Peptide receptor radionuclide therapy with 177 Lu-DOTATATE: the IEO phase I-II study. European journal of nuclear medicine and molecular imaging, 38(12), 2125-2135.
Brabander, T., Van der Zwan, W. A., Teunissen, J. J., Kam, B. L., Feelders, R. A., de Herder, W. W., Kwekkeboom, D. J. (2017). Long-term efficacy, survival, and safety of [177Lu-DOTA0, Tyr3]
octreotate in patients with gastroenteropancreatic and bronchial neuroendocrine tumors. Clinical Cancer Research, 23(16), 4617-4624.
Briggs, A. (2001). Handling uncertainty in economic evaluation and presenting the results. Economic evaluation in health care: merging theory with practice, 172-214.
Briggs, A., & Sculpher, M. (1998). An introduction to Markov modelling for economic evaluation. Pharmacoeconomics, 13(4), 397-409.
Briggs, A., Sculpher, M., & Claxton, K. (2006). Decision modelling for health economic evaluation: OUP Oxford.
Brouwer, W. B., Niessen, L. W., Postma, M. J., & Rutten, F. F. (2005). Need for differential discounting of costs and health effects in cost effectiveness analyses. Bmj, 331(7514), 446-448. Byford, S., & Raftery, J. (1998). Perspectives in economic evaluation. Bmj, 316(7143), 1529-1530. Casciano, R., Chulikavit, M., Perrin, A., Liu, Z., Wang, X., & Garrison, L. P. (2012). Cost-effectiveness of everolimus vs sunitinib in treating patients with advanced, progressive pancreatic neuroendocrine tumors in the United States. Journal of medical economics, 15(sup1), 55-64.
15
Ezziddin, S., Khalaf, F., Vanezi, M., Haslerud, T., Mayer, K., Al Zreiqat, A., Sabet, A. (2014). Outcome of peptide receptor radionuclide therapy with 177 Lu-octreotate in advanced grade 1/2 pancreatic neuroendocrine tumours. European journal of nuclear medicine and molecular imaging, 41(5), 925-933.
Fenwick, E., Claxton, K., & Sculpher, M. (2001). Representing uncertainty: the role of cost‐ effectiveness acceptability curves. Health economics, 10(8), 779-787.
Gravelle, H., & Smith, D. (2001). Discounting for health effects in cost–benefit and cost‐effectiveness analysis. Health economics, 10(7), 587-599.
Hakkaart-van Roijen, L., Van der Linden, N., Bouwmans, C., Kanters, T., & Tan, S. S. (2015). Kostenhandleiding. Methodologie van kostenonderzoek en referentieprijzen voor economische evaluaties in de gezondheidszorg. In opdracht van Zorginstituut Nederland. Geactualiseerde versie. Hordijk, L. (2019) Reconstructie Lutetium-octreotaat. Nederlands tijdschrift voor Geneeskunde. 2019;163: D3744.
Korse, C. M., Taal, B. G., van Velthuysen, M.-L. F., & Visser, O. (2013). Incidence and survival of neuroendocrine tumours in the Netherlands according to histological grade: experience of two decades of cancer registry. European journal of cancer, 49(8), 1975-1983.
Kwekkeboom, D. J., de Herder, W. W., Kam, B. L., van Eijck, C. H., van Essen, M., Kooij, P. P.,
Krenning, E. P. (2008). Treatment With the Radiolabeled Somatostatin Analog [^ 1^ 7^ 7Lu-DOTA^ 0, Tyr^ 3] Octreotate: Toxicity, Efficacy, and Survival. Journal of Clinical Oncology, 26(13), 2124-2130. Kwekkeboom, D. J., Teunissen, J. J., Bakker, W. H., Kooij, P. P., de Herder, W. W., Feelders, R. A., Krenning, E. P. (2005). Radiolabeled somatostatin analog [177Lu-DOTA0, Tyr3] octreotate in patients with endocrine gastroenteropancreatic tumors. Journal of Clinical Oncology, 23(12), 2754-2762. Limwattananon, S. (2011). Handling uncertainty of the economic evaluation result: sensitivity analysis. Journal of the Medical Association of Thailand, 91(6), 59.
Mujica-Mota, R., Varley-Campbell, J., Tikhonova, I., Cooper, C., Griffin, E., Haasova, M., Long, L. (2018). Everolimus, lutetium-177 DOTATATE and sunitinib for advanced, unresectable or metastatic neuroendocrine tumours with disease progression: a systematic review and cost-effectiveness analysis. Health Technology Assessment, 22(49).
NICE. (2008). Guide to the methods of technology appraisal. Londen: NICE, 2008. NU.nl/ANP (2019, January 9). Zwitsers bedrijf verzesvoudigt prijs zeldzaam Nederlands kankermedicijn. Retrieved from: nu.nl/binnenland/5672681 [Accessed January 20, 2019]. Raymond, E., Dahan, L., Raoul, J.-L., Bang, Y.-J., Borbath, I., Lombard-Bohas, C., Vinik, A. (2011). Sunitinib malate for the treatment of pancreatic neuroendocrine tumors. New England Journal of Medicine, 364(6), 501-513.
16
Sculpher, M., & Drummond, M. (2001). The role and estimation of productivity costs in economic evaluation. Economic evaluation in health care: merging theory with practice, 94-112.
Sheldon, T. (2012). Dutch doctors call for EU evaluation of cost effectiveness of high cost orphan drugs. In: British Medical Journal Publishing Group.
Strosberg, J., El-Haddad, G., Wolin, E., Hendifar, A., Yao, J., Chasen, B., Jacene, H. (2017). Phase 3 trial of 177Lu-Dotatate for midgut neuroendocrine tumors. New England Journal of Medicine, 376(2), 125-135.
Strosberg, J., Wolin, E., Chasen, B., Kulke, M., Bushnell, D., Caplin, M., Hendifar, A. (2018). Health-related quality of life in patients with progressive midgut neuroendocrine tumors treated with 177Lu-Dotatate in the phase III NETTER-1 trial. Journal of Clinical Oncology, 36(25), 2578.
Strosberg, J. R., Wolin, E. M., Chasen, B. A., Kulke, M. H., Bushnell, D. L., Caplin, M. E., Lopera Sierra, M. (2018). First update on overall survival, progression-free survival, and health-related time-to-deterioration quality of life from the NETTER-1 study: 177Lu-Dotatate vs. high dose octreotide in progressive midgut neuroendocrine tumors. In: American Society of Clinical Oncology.
Swinburn, P., Wang, J., Chandiwana, D., Mansoor, W., & Lloyd, A. (2012). Elicitation of health state utilities in neuroendocrine tumours. Journal of medical economics, 15(4), 681-687.
Tan, S. S., Bouwmans, C. A., Rutten, F. F., & Hakkaart-van Roijen, L. (2012). Update of the Dutch manual for costing in economic evaluations. International journal of technology assessment in health care, 28(2), 152-158.
Trotti, A., Colevas, A. D., Setser, A., Rusch, V., Jaques, D., Budach, V., Coleman, C. N. (2003). CTCAE v3. 0: development of a comprehensive grading system for the adverse effects of cancer treatment. Paper presented at the Seminars in radiation oncology.
van Baal, P. H., Wong, A., Slobbe, L. C., Polder, J. J., Brouwer, W. B., & de Wit, G. A. (2011).
Standardizing the inclusion of indirect medical costs in economic evaluations. Pharmacoeconomics, 29(3), 175-187.
Volksgezonheidenzorg.info (2017) Sterftecijfers huidige situatie. URL:
volksgezondheidenzorg.info/onderwerp/sterfte/cijfers-context/huidige-situatie [Accessed May 14, 2019]
Yao, J. C., Shah, M. H., Ito, T., Bohas, C. L., Wolin, E. M., Van Cutsem, E., De Vries, E. G. (2011). Everolimus for advanced pancreatic neuroendocrine tumors. New England Journal of Medicine, 364(6), 514-523.
Zorginstituut Nederland URL: https://medicijnkosten.nl [Accessed May 15, 2019]
Zwaap, J., Knies, S., Van der Meijden, C., Staal, P., & Van der Heiden, L. (2015). Kosteneffectiviteit in de praktijk. Diemen: Zorginstituut Nederland.
