Cost-effectiveness analysis of pancreatin minimicrospheres in patients with pancreatic exocrine insufficiency due to chronic pancreatitis

Abstract

Objective:

Chronic pancreatitis (CP) is the most common cause of pancreatic exocrine insufficiency (PEI). Management of PEI due to CP is achieved through lifelong treatment with pancreatic enzyme replacement therapy (PERT). To the authors’ knowledge, no cost-effectiveness analysis on the benefit of PERT in CP patients with PEI has been performed to date. The objective of this analysis was to examine the cost-effectiveness of Creon (pancreatin minimicrospheres [MMS]), one of the main PERTs available in Poland, in treating patients with CP-related PEI.

Methods:

The cost-effectiveness of pancreatin MMS in the treatment of patients with CP-related PEI vs no PERT treatment was estimated using a decision analysis based on clinical data from relevant studies. The model horizon was 20 years. Main outcomes included the percentage of patients with controlled PEI, survival, total medical costs, number of quality-adjusted life years (QALYs), and the incremental cost-effectiveness ratio (ICER). All costs were analysed from the Polish payer perspective.

Results:

The model included clinical data from 176 patients treated in five pancreatin MMS randomized trials. Treatment with pancreatin MMS resulted in a considerably higher proportion of patients with controlled PEI compared to those not treated with any PERT. Over a horizon of 20 years, the total treatment cost and the ICER for pancreatin MMS was €8223 and €6312 per QALY, respectively.

Limitations:

Important limitations include the lack of long-term and comparative clinical data available. The use of ‘no PERT treatment’ as a comparator against pancreatin MMS treatment may not accurately reflect current practice in Poland.

Conclusions:

Treatment of CP-related PEI with pancreatin MMS is cost-effective from a Polish payer perspective, with an ICER below the accepted ‘willingness to pay’ threshold of 3-times gross domestic product (GDP) per capita. These results are likely to apply to other European countries.

Keywords::

• Chronic pancreatitis
• Pancreatic exocrine insufficiency
• Pancreatin
• Cost-effectiveness analysis
• Cost utility analysis
• Health economic model

Abbreviations::

• AHTAPol, Agency for Health Technology Assessment in Poland; CFA, Coefficient of fat absorption; CI, Confidence interval; CP, Chronic pancreatitis; GDP, Gross domestic product; HRQoL, Health-related quality-of-life; ICER, Incremental cost-effective ratio; LOCF, Last observation carried forward; MMS, Minimicrosphere; P, Placebo; PEI, Pancreatic exocrine insufficiency; PERT, Pancreatic enzyme replacement therapy; PMM, Pancreatin minimicrospheres; PSA, Probabilistic sensitivity analysis; QALY, Quality-adjusted life year; QoL, Quality-of-life; RCT, Randomized clinical trial; SA, Sensitivity analysis; SD, Standard deviation

Introduction

Pancreatic exocrine insufficiency (PEI) is the failure of the pancreatic exocrine function as a result of obstruction of the main pancreatic duct, loss of the pancreatic parenchyma, decreased pancreatic stimulation, or acid-mediated inactivation of pancreatic enzymes. It typically presents when post-prandial secretion of pancreatic enzymes falls below 10% of normal levels1,2. As a consequence, fat malabsorption and maldigestion takes place resulting in, amongst others, malnutrition, weight loss, abdominal pain, flatulence, and steatorrhoea. Major causes of PEI include underlying diseases such as cystic fibrosis, pancreatic cancer, chronic pancreatitis (CP), pancreatic surgery, and trauma to the pancreas. Of these, CP is the most common cause of PEI3.

Patients with PEI due to CP have substantially impaired quality-of-life (QoL)4,5. Primary complaints include severe diarrhoea, abdominal pain, and malnutrition which sometimes require multiple hospitalizations6. This can significantly affect a patient’s daily function with some no longer being able to continue with their job7,8. In practice, diagnosis focuses largely on characteristic symptoms of PEI such as steatorrhea and malnutrition7. Clinical testing to measure a patient’s coefficient of fat absorption (CFA) is a more reliable method of diagnosis, but is mostly reserved for clinical trials. CFA is the percentage measure of fat absorbed by the body compared to the amount of fat ingested in grams9. Stool fat excretion can also be used for diagnosis of PEI. However in this case the outcome is dependent on the fat intake of patients, making comparisons between patient results difficult if fat intake is not standardized. Another diagnostic method recommended by a number of clinical guidelines is to measure the level of elastase 1 activity in patients’ faeces10–12. Elastase 1 activity below 200 μg/g is indicative of mild PEI whereas levels below 50–100 μg/g represent patients with severe PEI12. However, faecal elastase is independent of pancreatic enzyme supplementation and is, therefore, not suitable as a clinical end-point. Because of these shortcomings, most physicians continue to base their diagnosis mainly on clinical symptoms7.

Treatment of PEI aims at symptom relief and at re-establishing the normal nutritional status in these patients13. This is achieved with pancreatic enzyme replacement therapies (PERTs) which improve digestion and absorption of fat and other nutrients.

The incidence of CP in Poland is on average five patients per 100,000 inhabitants per year12,14. This figure ranges from 6–8.6 CP patients per 100,000 per year across Europe15–17. The measurement of CP prevalence is a much more difficult task and studies that assess prevalence rates should be regarded with caution; frequently prevalence is underestimated. While new patients with fresh CP symptoms are likely to see a gastroenterologist (therefore incidence is likely to be assessed correctly), at a later stage of their disease CP patients are often treated by their General Practitioner or not seeing a doctor at all. Prevalence of PEI in CP is difficult to assess and can vary greatly due to the occult nature of the illness, especially during earlier stages. This can be seen with figures from a German and a Belgian study, which report a prevalence of 11.5% and 63%, respectively18,19. In the Belgian study, this value accounts for patients with CP for at least 5 years. In the same study, 94% of CP patients had developed PEI after 10 years of onset. No recent data on the prevalence of PEI due to CP in Poland seems to have been published. However, the Agency for Health Technology Assessment in Poland (AHTAPol) applies a rate of 60% following a national consultant in gastroenterology20. According to recently updated Polish clinical guidelines, alcohol is undoubtedly a risk factor, but many other factors are still unknown12. Several observational studies conducted during the 1960s to the 1990s reported that chronic alcohol abuse was considered the primary risk factor of CP (60–90%)21–28. However, more recent multi-centre studies reported a lower proportion of patients in whom alcohol was considered a single or contributing aetiologic factor (∼ 40–50%)29–33. In addition to this, it is becoming more evident that smoking is also an important risk factor of CP34. Of the 34 physicians who provided input to the latest Polish clinical guideline, all agreed that cessation of smoking can slow the progression of CP12. Data on the effects of alcohol abstinence on disease progression remain inconclusive10,35.

A recent Polish study reported that 44% of patients with CP retire early from employment due to deterioration of QoL as a result of CP progression8. In the same study, low general health scores were strongly correlated with both PEI and unemployment. Overall, 64% of CP patients below the age of 50 were either retired or unemployed. In relation to these figures, there is concern that many patients cannot afford treatment or continue their treatment over a long period of time7. Both Ockenga36 and Dumasy19 report that CP patients with untreated PEI suffer from higher levels of malnutrition, significantly greater weight loss, and increased mortality and morbidity compared to CP patients who are sufficiently treated with PERT. During the time the present study was carried out, no PERT indicated for CP-related PEI was on the drug reimbursement list in Poland. Only one reimbursement submission had been made to AHTAPol prior to this study37. Here pancreatin was not recommended for reimbursement due to lack of clinical data submitted showing effectiveness in the proposed indication. Reimbursement of PERTs in Poland could lead to better management of these patients.

To our knowledge no cost-effectiveness analysis assessing the benefit of PERT in CP patients with PEI has been performed to date. The objective of this analysis was to assess the cost-effectiveness of Creon (pancreatin MMS; Abbott Products GmbH, Hannover, Germany) for the treatment of CP-related PEI vs no PERT treatment from the Polish payer perspective. Pancreatin enteric-coated MMS is a PERT indicated for the treatment of PEI due to CP and other conditions including, but not limited to, cystic fibrosis, pancreatic cancer, and gastrointestinal surgery (particularly pancreatectomy)38. This formulation of pancreatin has been shown to be effective in improving fat malabsorption and maldigestion in patients with CP-related PEI39–43.

Methods

Model structure

The cost-effectiveness of pancreatin MMS for the treatment of CP-related PEI was estimated using a decision analysis model. Pancreatin MMS treatment was modelled against no PERT treatment as there was insufficient data on other PERTs prescribed in Poland. The health economic model was based on clinical data from five pancreatin MMS randomized clinical trials (RCTs), published literature, and expert interviews.

A Markov model structure (Figure 1) was developed based on the following health states: uncontrolled PEI, controlled PEI, and death. In line with the clinical studies where CFA was the primary end-point, CFA was the main outcome of the model. Controlled PEI was defined as having a CFA above 80% and uncontrolled PEI as having a CFA equal to or below 80%7. Patients with a CFA above 80% were associated with a healthier nutritional state (i.e. controlled PEI). Appling this threshold to the model was considered as a reasonable hypothesis by a group of European clinical experts7.

Figure 1.  Core model structure (right) and treatment algorithm (left) for patients with uncontrolled PEI in the pancreatin MMS treatment arm.

As CP-related PEI is a chronic disease without a cure, the model’s time horizon should cover a patient’s life time. Reported mortality at 20 years ranges between 37–62%23,44,45. We assume a 55% mortality rate after 20 years of disease. For the purpose of this article, a time horizon of 20 years was applied, as mortality data was available only for a period of 20 years. The initial cycle length of the model was 2 weeks, with subsequent cycles of 6 months.

At the beginning of the model all patients are considered to be in the uncontrolled PEI state (i.e. CFA ≤ 80%) and receive either treatment with pancreatin MMS oral capsules or no PERT treatment (Figure 1). Pancreatin MMS contains pancreatic enzymes amylase, lipase, and protease, and is available in a number of different formulations and strengths to meet the needs of all age groups and allow for individual dosing. The initial pancreatin MMS regimen used in the model was 40,000 lipase units per main meal (3-times per day) and 25,000 lipase units per snack (2-times per day). The dosing regimen applied to the model corresponds to the overall average amount administered in the trials and was validated with clinical experts7. These doses are considered high compared to the latest Polish clinical guideline which recommends 25,000–40,000 lipase units per main meal and 20,000 lipase units per snack12. On the contrary, some key opinion leaders argue that the starting dose should be higher (minimum 40,000–50,000 lipase units per main meal and 10,000–25,000 lipase units per snack) to guarantee optimal efficacy of PERT7,40,46–48.

At the end of the initial model cycle (2 weeks), patients may either become controlled (CFA > 80%) or remain uncontrolled (CFA ≤ 80%). No patient is assumed to die in this cycle. Patients treated with pancreatin MMS who are controlled after the first 2 weeks are prescribed the same treatment and dose for the remainder of the model. Patients with uncontrolled PEI after the initial 2 weeks remain on the initial pancreatin MMS regimen for the next model cycle (6 months). If still uncontrolled after the first 6 months, the pancreatin MMS dose is doubled, in line with expert opinion7, and patients enter the increased dose health state (Figure 1). These patients receive the increased dose for the remainder of the model unless the patient discontinues pancreatin MMS or dies. This assumption was taken in line with the conservative approach of the model; however, it may not reflect real life clinical practice where these patients are likely to have their dose increased much sooner7. Efficacy of the increased pancreatin MMS dose regimen was assumed to be the same as for the initial dose regimen. Patients who remain uncontrolled after 12 months of treatment are assumed to discontinue treatment and remain in this uncontrolled PEI health state until they die.

Main outcomes estimated for each treatment arm included the percentage of patients with controlled PEI, survival, PERT treatment cost, the number of QALYs, and the incremental cost-effectiveness ratio (ICER). All costs and health outcomes are presented as per-person averages. To determine the ICER of treatment with pancreatin MMS over the no PERT treatment arm a roll-back analysis (i.e. averaging out and fold backing the tree) was performed.

Model population

For the model analysis, the population included in this analysis consisted of adult patients with CP-related PEI who were treated with either pancreatin MMS or received no PERT treatment. Data for the treatment effect of pancreatin MMS on CFA was derived from five RCTs (Table 1)39–43. The placebo results from these trials, when available, were used to determine the no PERT treatment effect and transition probabilities. Patients who had undergone pancreatic surgery were excluded from the model because of very limited available data. In total, data from 176 patients were considered for the analysis. The average baseline age for this patient population was 49 years (range, 18–75). Any missing data was imputed using the last observation carried forward (LOCF) method. All five RCTs were conducted in accordance with the Good Clinical Practice Guidelines and the ethical principles laid down in the Declaration of Helsinki.

Table 1.  Treatment effect from identified clinical trials.

Reference	Countries	Treatment duration (days)	Number of patients per treatment arm	Mean fat excretion (g/day)^a	CFA (%)^a
Safdi et al.40	US	14	PMM = 12	PMM = 22.9	PMM = 86.6
			P = 14	P = 51.8	P = 68.00
Halm et al.41^b	Germany	14	PMM = 11	PMM = 17.3	PMM = 77.2
O’Keefe et al.39	South Africa	14	PMM = 15	PMM = 19.6	PMM = 81.5
			P = 16	P = 43.3	P = 60.3
Whitcomb et al.42	Bulgaria, Poland, Russia, Serbia, Ukraine, US	7	PMM = 24	PMM = 67.3	PMM = 86.1
			P = 28	P = 165.8	P = 66.90
Thorat et al.43	India	7	PMM = 32	PMM = 16.2	PMM = 86.1
			P = 24	P = 30.0	P = 72.9
		365	PMM = 47	PMM = 12.2	PMM = 88.91

CFA, Coefficient of fat absorption; P, Placebo treatment group; PMM, Pancreatin minimicrospheres treatment group.

^aThe reported treatment effect refers to the assessment at the end of the treatment duration.

^bOnly data from the first treatment period from the MMS treatment arm were included due to the model design. Only patients with a post-efficacy measurement were taken into account.

All studies identified for the model were multi-centre, double-blind, and placebo-controlled40–43, with exception of one cross-over study39 which compared two different formulations of pancreatin MMS. Pancreatin MMS data for the short-term treatment effect were derived from all five studies. Long-term treatment effect data was only available from one of the five studies43. This particular study comprised of a 1-week double-blind, randomized, placebo controlled period followed by an open-label long-term extension period of nearly 1 year43. Only patients with PEI due to CP, defined as stool fat excretion ≥10 g/day and a CFA < 80%, were included in these studies. All RCTs used mean CFA or change in CFA from baseline as their primary outcome. Treatment doses ranged from 100,000–400,000 lipase units per day. Two of the trials reported CFA at week 142,43 and the others at week 239–41. As the treatment effect at week 2 was required for the model, it was decided to use the pooled data from all five studies and assume this as the effectiveness at week 2 as the efficacy of pancreatin MMS in both short-term periods (i.e. week 1 and week 2) was comparable. Results from all RCTs showed a clear difference between treatment groups in that mean stool fat excretion per day was reported to be far lower in the pancreatin MMS treatment group compared to placebo (Table 1). Compared to patients treated with placebo, pancreatin MMS-treated patients achieved a higher CFA. The overall mean CFA reported for the pancreatin MMS treatment group, taking into account results from all the five selected RCTs, ranged from 77.2–88.9%39–43. The mean CFA for the placebo group in the five studies ranged from 60.3–72.9%. The highest mean CFA achieved under pancreatin MMS treatment was reported in the long-term extension study (88.9%)43.

The probability of CP patients transitioning from one health state to another for the short-term treatment effect (first cycle, Table 2) was determined by comparing the number of uncontrolled patients at baseline to the number of controlled and uncontrolled patients at the end of treatment using results from the short-term clinical studies mentioned above39–43. The same method was applied to determine the transition probability for the long-term treatment effect using data from the long-term extension study43. As all patients were treated with pancreatin MMS during this extension period no data was available for placebo long-term efficacy. Additional literature research on long-term data did not reveal any further insights in the efficacy of no PERT treatment vs pancreatin MMS in patients with PEI due to CP. It was, therefore, assumed that patients who are controlled after 2 weeks of no PERT treatment have the same chance of remaining controlled for the remaining cycles as those who are treated with pancreatin MMS. With CP being a progressive disease, this is a very conservative assumption.

Table 2.  Model input parameters.

Parameter	Base case value	Values used in SA (range)	PSA distribution (PSA parameters)		Source
Model Horizon
Discount rate for costs	5.0%	0–5%	None		ATOM51
Discount rate for outcomes	3.5%	0–5%	None
Utilities
Mild pancreatitis (controlled PEI)	0.95	0.85–0.95	Beta	19.02; 2.11	Poulose et al.49
Severe pancreatitis (uncontrolled PEI)	0.71	0.18–0.71	Beta	24.25; 9.91
Mortality
5-year death rate	15%	3–18%	Beta	235 (1565)	Lowenfels et al.44
10-year death rate	30%	14–34%	Beta	292 (973)
15-year death rate	45%	20–48%	Beta	246 (547)
20-year death rate	55%	37–62%	Beta	112 (204)
Dosing Regimen of Pancreatin MMS
Time at which pts receive increased dose	2 weeks	2–52	None		Clinical experts7
% pts with increased dosing	100%	0–100%	None
Dose increase rate	2	0–3	None
Time at which pts discontinue	26 weeks	2–260	None		Assumption
% pts who discontinue	100%	0–100%	None		Assumption
Medication Costs
Conversion rate Złoty to Euro	0.2413	0.217–0.265	None		Oanda54
Cost per capsule (Pancreatin MMS 40,000 lipase units)	€0.61	–	None		Provided by Abbott
Cost per capsule (Pancreatin MMS 25,000 lipase units)	€0.28	–	None		Wykaz Leków Refundowanych50
Transition Probabilities
First cycle*					O’Keefe et al.39
No PERT treatment: Uncontrolled to controlled	27%	18–37%	Beta	21 (79)	Safdi et al.40 Halm et al.41
Pancreatin MMS: Uncontrolled to controlled	69%	59–79%	Beta	59 (85)	Whitcomb et al.42 Thorat et al.43
Next cycles					Thorat et al.43
No PERT treatment: Remaining in controlled	100%	81–100%	Beta	27 (27)
Pancreatin MMS: Remaining in controlled	100%	81–100%	Beta	27 (27)
No PERT treatment: Uncontrolled to controlled	0%	0–10%	None
Pancreatin MMS: Uncontrolled to controlled	35%	12–66%	Beta	4 (7)

MMS, Minimicrospheres; PEI, Pancreatic exocrine insufficiency; PERT, Pancreatic enzyme replacement therapy; PSA, Probabilistic sensitivity analysis; SA, Sensitivity analysis.

*For the short-term treatment effect, the table only shows the per-cycle probability to move from one health state to another. The chance of remaining in the present health state is 1 minus the transition probability of moving away.

Overall, the RCTs reported that pancreatin MMS was generally well-tolerated and had a similar adverse event profile to that of placebo in patients with CP-related PEI39–43. Reported treatment-emergent adverse events from these trials were low in number and were primarily gastrointestinal events (mainly diarrhoea and constipation) and metabolic/nutritional disorders. Most trials reported no PERT treatment-related serious adverse events. No patient withdrawals occurred in either treatment group as a result of an adverse event. Hence, adverse event data was not included in the analysis.

Utilities and health resource utilisation

No studies were identified that reported any utilities or mapped health-related quality-of-life (HRQoL) parameters for patients with PEI. A study by Poulose et al.49 provided an estimate on utility for patients with mild and severe pancreatitis (Table 2). In our model it is assumed that patients with uncontrolled PEI have the same HRQoL as patients with severe pancreatitis, and that patients with controlled PEI have the same HRQoL as patients with mild pancreatitis. This was confirmed by clinical experts7. When asked for their own rating of QoL, they estimate the utility score of patients with controlled PEI to be an average of 85 (range, 80–100) and uncontrolled PEI to be an average of 17.5 (range, 0–30), based on a scale from 0–100, where 0 corresponds to the worst HRQoL. These values were applied in the sensitivity analyses.

Only medical costs (i.e. treatment cost of pancreatin MMS) were included in the analysis, as very limited cost data were available. These costs were based on 2011 values extracted from the Polish pricing book for drugs50 and data provided by Abbott (Table 2). No other direct costs or any indirect costs have been incorporated in the model as they were not available. Costs were discounted at a constant rate of 5% and health outcomes were discounted at a constant rate of 3.5% according to Polish guidelines on conducting pharmacoeconomic evaluations51.

Mortality data applied to the model was derived from a large scale study with long follow-up on European CP patients who did not have pancreatic surgery44 (Table 2). For the analysis, mortality is assumed to be the same for each health state and only differs in time.

Additional analyses

Further analyses were performed including an extensive one-way sensitivity analysis (SA), a probabilistic sensitivity analysis (PSA), and a scenario analysis to evaluate the parameters to which the model results are most sensitive and to test the robustness of the cost-effectiveness results. All input parameters are provided in Table 2.

Results

Table 3 provides a summary of the main outcome measures of the base case analysis after 20 years. Results show that, after 20 years, 41% of patients on pancreatin MMS had controlled PEI compared to 12% on no PERT treatment. The percentage of patients on pancreatin MMS with uncontrolled PEI (4%) was considerably lower than those who received no PERT treatment (33%). As there were no costs associated with no PERT treatment, the cost difference between the two treatment arms was equal to the total cost of pancreatin MMS treatment (20 years: €8223/patient). Over the 20-years horizon, treatment of CP-related PEI with pancreatin MMS resulted in a gain of 1.31 QALYs compared to no PERT treatment (9.45 vs 8.14). The ICER for pancreatin MMS treatment was €6312, which is below the willingness-to-pay threshold applied by the Polish health authorities (i.e. less than 3-times the set gross domestic product (GDP) per capita; 33,181 PLN × 3 = 99,543 PLN = 24,148 EUR [21 March 2012])52–54. The GDP per capita estimate is a set value based on 2007–2009 Polish data and is utilized by the Polish health authorities effectively until the 31st of November 201253.

Table 3.  Results for 5-, 10-, and 20-year cost-effectiveness analysis.

	SA, 5 years		SA, 10 years		Base case analysis, 20 years
Treatment arm	Pancreatin MMS	No PERT treatment	Pancreatin MMS	No PERT treatment	Pancreatin MMS	No PERT treatment
Controlled PEI	78%	23%	64%	19%	41%	12%
Uncontrolled PEI	7%	62%	6%	51%	4%	33%
Died	15%	15%	30%	30%	55%	55%
Costs (€)	3473	0	5816	0	8223	0
QALYs	3.64	3.15	6.36	5.49	9.45	8.14
Cost (€) difference	3473		5816		8223
QALY difference	0.49		0.87		1.31
ICER (€)	7036		6666		6312

ICER, Incremental cost-effective ratio; MMS, Minimicrospheres; PEI, Pancreatic exocrine insufficiency; QALY, Quality-Adjusted Life Year; SA, Sensitivity analysis.

Sensitivity analysis results

A SA was performed using shorter time horizons of 5 and 10 years. Results for this SA are presented in Table 3 aside the base case analysis results. After 5 years of treatment, the proportion of patients with controlled PEI compared to uncontrolled was 78% and 7%, respectively. Of those receiving no PERT treatment, 23% had controlled PEI and 62% had uncontrolled PEI. The mortality rate after 5 years was 15% for both treatment groups. Over the 5 year horizon, treatment with pancreatin MMS resulted in an ICER of €7036.

After 10 years, 64% of patients treated with pancreatin MMS had controlled PEI, whereas 6% of treated patients were uncontrolled. Of patients who received no PERT treatment, 19% had controlled PEI and 51% had uncontrolled PEI. Both treatment groups had a mortality rate of 30%. The ICER for treatment with pancreatin MMS after 10 years was €6666.

Table 4 shows the impact of the input parameters on the ICER of pancreatin MMS treatment compared to the no PERT treatment alternative. Results from this SA show that six parameters have a strong influence on the ICER after 20 years of treatment, while the remaining input parameters had only little influence on the ICER. Specifically, the long-term transition probability of pancreatin MMS patients remaining controlled and the probability of uncontrolled patients receiving no PERT treatment becoming controlled after 1 year had the largest impact on the ICER. When the probability of becoming controlled after 1 year would reduce from 100% to 81%, then pancreatin MMS treatment would be dominated (i.e. more costly and less effective). Other highly influential parameters included utilities for controlled and uncontrolled CP patients, the cost conversion rate, and the transition probability of patients in the no PERT treatment group moving from uncontrolled to controlled after 1 year and remaining controlled thereafter.

Table 4.  One-way sensitivity analysis results.

Parameter	SA value	ICER (€)
	Min	Max	Run 1	Run 2
Base case	–	–	6,312	6,312
Pancreatin MMS; Controlled to controlled after 1 year	81%	100%	−84,299	6,312
No PERT treatment; Uncontrolled to controlled after 1 year	0%	10%	6,312	28,422
Utility uncontrolled PEI	0.18	0.71	1,654	6,312
Utility controlled PEI	0.85	0.95	8,567	4,997
No PERT treatment; Controlled to controlled after 1 year	81%	100%	4,704	6,312
Conversion rate	0.208	0.255	5,681	6,944
Dose increase (same for Pancreatin MMS 40,000 and 25,000 lipase units)	0	3	5,700	6,619
Pancreatin MMS; Uncontrolled to controlled after 1 year	12%	66%	6,801	5,892
Time at which uncontrolled patients receive increased dose	2	52	6,804	6,006
Time after which uncontrolled Pancreatin MMS patients discontinue	2	260	6,250	6,609
% uncontrolled patients receiving increased dose after 26 weeks	0%	100%	6,006	6,312
% uncontrolled patients discontinuing Pancreatin MMS after 52 weeks	0%	100%	6,614	6,312
Mortality	37%	62%	6,237	6,334
Discounting on cost and outcome	0%	5%	6,918	6,970

ICER, incremental cost-effectiveness ratio; MMS, Minimicrospheres; PEI, Pancreatic exocrine insufficiency; PERT, Pancreatic enzyme replacement therapy; SA, Sensitivity analysis.

The minimum (Min) and maximum (Max) values used for each model input parameter in the SA as well as the resulting ICERs of the analysis are provided in the table. Run 1 ICER results correspond to minimum value input parameters whereas Run 2 outcomes correspond to maximum value input parameters. Parameters are ordered from most influential (top) to least influential (bottom) based on SA results.

Probabilistic sensitivity analysis results

Results of the PSA are presented in Table 5. The mean incremental cost after 10,000 simulations was €8025 (95% confidence interval [CI] €5830–€8652), with a mean incremental benefit of 1.43 QALYs (95% CI = 1.08–1.67). The mean ICER after 10,000 simulations was €13,425, with a median value of €5659. The difference between the median and mean is caused by some outliers in the QALY difference. The probability of pancreatin MMS being cost-effective compared to no PERT treatment was 96.3%, assuming a willingness-to-pay limit of €20,000 per QALY gained.

Table 5.  Probabilistic sensitivity analysis results.

Result	Incremental cost (€)	Incremental QALYs	ICER (€)
Base case	8223	1.30	6,312
PSA mean	8025	1.43	13,425
PSA median	8392	1.39	5,659
SD	769	0.81	115,363
95% CI	5830–8652	1.08–1.67	Dominated – 2,258
Min/max	4390–8809	−2.07–4.61	Dominated – 1,824

CI, confidence interval; MMS, Minimicrospheres; QALY, Quality-Adjusted Life Year; PSA, Probabilistic sensitivity analysis; SD, Standard deviation.

Scenario analysis

In the scenario analysis, CFA was replaced with stool fat excretion as the base case clinical outcome. In this case, uncontrolled PEI due to CP was defined as stool fat excretion greater than 10 g/day. Results are summarized in Table 6. This change of clinical outcome had a great impact on treatment efficacy, with 0% of controlled patients in both treatment arms after 20 years. After 20 years patients were either uncontrolled (45%) or dead (55%). The difference in QALYs was reduced to 0.09, but treatment with pancreatin MMS remained cost-effective, with an ICER of €17,443.

Table 6.  Scenario analysis results based on stool fat excretion below 10 g/day.

Stool fat below 10 g/day	Pancreatin MMS	No PERT treatment
Controlled PEI	0%	0%
Uncontrolled PEI	45%	45%
Dead Patients	55%	55%
Costs (€)	1,616	0
QALYs	7.70	7.61
Cost (€) difference	1,616
QALY difference	0.09
ICER	17,443

ICER, Incremental cost-effective ratio; MMS, Minimicrospheres; PEI, Pancreatic exocrine insufficiency; QALY, Quality-Adjusted Life Year.

Controlled CP-related PEI was defined as stool fat excretion < 10 g/day.

Discussion

This is the first study that compares the cost-effectiveness of treating patients with CP-related PEI with pancreatin MMS vs not treating these patients with any PERT. The economic evaluation described here shows that PERT treatment with pancreatin MMS in patients with CP-related PEI is cost-effective from a Polish payer perspective. Considering that a large proportion of efficacy data used in the model originates from European patients, it seems legitimate to extrapolate that pancreatin MMS would be cost-effective throughout Europe. This is particularly true since only costs of pancreatin MMS acquisition were used in the model. The cost of pancreatin MMS is relatively similar in many countries across Europe. In Western Europe, pancreatin MMS is reimbursed for treatment of CP-related PEI in several regions and countries including the UK, Nordics, France, Benelux, Germany, Austria, Switzerland, Spain, Portugal, and Italy.

A large proportion of the clinical efficacy data applied to the model originated from four short-term studies (with exclusion of a 1 year Indian study43). Clinical experts have stated that these studies were sufficient for drug approval as a treatment effect is already evident after 2 weeks7. It addition to this, pancreatin has effectively been used for treatment of PEI for many years. Pancreatin products have been marketed by Abbott (former Solvay Pharmaceuticals) for over 100 years. The drug was first approved on 14 April 1980 in Austria for the treatment of PEI in paediatric and adult patients. Although older pancreatin registrations from as early as 1920 exist, since its introduction to the market, numerous strengths and formulations of pancreatin for the treatment of PEI have been authorised around the globe.

CP is the main cause of PEI. In the Western population, alcohol abuse is the common cause of CP55. Patients with PEI who experience weight loss or present relevant steatorrhea and steatorrhea-related symptoms are classically considered to be suitable candidates for PERT. These patients consistently suffer from malnutrition. Prescription of PERT prevents potentially relevant nutritional deficits. Recently published Polish clinical practice guidelines for the treatment of CP recommend patients with symptomatic PEI to be treated with PERT12. PERT is to be maintained throughout life. While the importance of treating these patients is recognized, these recommendations are not always implemented in real life clinical practice, probably because no PERT for the treatment of CP-related PEI was reimbursed in Poland. Consequently, many patients were at risk of poor disease maintenance.

The base case analysis demonstrated that pancreatin MMS is a cost-effective treatment with an ICER of €6312. These results were robust within the parameter ranges tested with 96% of the simulations being below the 3-times GDP per capita threshold. The sensitivity analyses indicate that there is much uncertainty around the long-term sustained effect of pancreatin MMS and this resulted in a large standard deviation in the reported ICERs. The one-way sensitivity analyses showed that the probability of remaining controlled on pancreatin MMS as well as the possibility to become controlled with no treatment after the initial 2-week period had a great impact on the ICER. As expected, the utilities allocated to controlled or uncontrolled PEI were important model drivers too, although the impact was much smaller than the long-term treatment effect on CFA. Short-term efficacy and mortality had little impact on the ICER.

Our model is associated with a number of limitations. As shown by the sensitivity analysis, the most important limitation is the lack of long-term data. The only available long-term data comes from a 52 week open-label extension phase of a 2-week RCT43. This extension phase study provided 1-year data for 32 patients out of the 34 who started on pancreatin MMS at the beginning of the short-term treatment period and completed the 1-year follow-up. This extension phase study also provided long-term data for patients who had been on placebo in the double-blind phase and were switched to pancreatin MMS in the open-label extension period for ethical reasons. Consequently, no long-term data is available for the ‘no treatment’ option and assumptions on the long-term effect for no treatment had to be made: (a) that the transitions observed between week 2 and week 52 would remain constant for the remainder of the model period and (b) that patients not receiving treatment would not be able to have their PEI controlled at a later point in time. While these assumptions seem reasonable and while the cost-effectiveness of pancreatin MMS at a short-term horizon of 2 weeks already leads to an acceptable ICER (€10,755), further long-term data on pancreatin MMS would make the model more robust.

The use of identical mortality rates at 5-, 10-, and 20-years time horizon (15%, 30%, and 55%, respectively) for both treatment groups could be considered as a limiting factor. These rates were based on data from three CP epidemiological studies23,44,45. One could expect that the mortality rates for patients receiving PERT treatment would be lower than that of patients receiving no PERT treatment. However, PERT does not treat CP but rather improves the QoL of patients by managing PEI. Clinical experts have confirmed that the underlying CP usually gradually worsens over a patients’ life, regardless of whether they receive PERT or not7. Thus no mortality differential was applied to the model. The main differential effect on QALYs in the model stemmed from the cumulative utility difference.

Another limitation identified is the use of ‘no PERT treatment’ as a comparator against pancreatin MMS treatment, which may not accurately reflect current practice in Poland. Indeed, there are other PERTs available on the Polish market but are far less frequently prescribed than pancreatin MMS. In the first half of 2012, pancreatin MMS accounted for roughly 2/3 of the entire Polish PERT market56. Of the PERTs currently reimbursed in Poland, pancreatin MMS has an average market share of 76.84%, whereas Lipancrea (pancreatin, Warszawa ZF Polfa) accounts for an average of 23.16% in 2012 so far57. With the lack of published clinical data on other PERTs and no PERT product being reimbursed for CP in Poland (i.e. at the time of the present study), ‘no treatment’ was used as the comparator as it was deemed impossible to compare pancreatin MMS to any other PERT.

The scenario analyses showed that the model is sensitive to the definition used for controlled PEI. When controlled PEI was defined by stool fat excretion rather than CFA, pancreatin MMS was still cost-effective with an ICER of €17,443, but the magnitude of its treatment effect decreased substantially. After 20 years, all patients in both treatment groups had uncontrolled PEI. Treatment with pancreatin MMS resulted in a gain of only 0.09 additional QALYs. Nevertheless, for understandable reasons neither CFA nor fat excretion are commonly used in clinical practice. Alternative methods to determine the effects of PERTs are needed in order to give a more real-life outcome measure.

No cost-effectiveness study has previously addressed the use of PERT in CP-related PEI. Similarly, there is no published economic data regarding treatment of CP or CP-related PEI in Poland. This rendered the pancreatin MMS economic model a challenge. In our model only direct treatment costs are considered. No other direct or any indirect costs are taken into account. Given the beneficial clinical outcomes for pancreatin MMS, including indirect costs which occur to society due to absenteeism from work may have improved the ICER further for pancreatin MMS. Insufficient data is available to support that treatment of PEI with PERT leads to fewer resource utilization. Thus, it has been assumed equal between the pancreatin MMS and no treatment arm. However, US hospital discharge statistics58 show that the average length of stay for patients with CP is significantly longer if malnutrition as a secondary diagnosis has been reported. In 2010, the average length of stay for patients with CP was 5.3 days, and in patients with CP and maldigestion it was 12.5 days. Similar findings were reported in a German study where the average length of stay for CP patients with nutritional risk was 9 days compared to CP patients without nutritional risk (6 days)36. When prescribed with dietary counselling, PERT reduces the extent of steatorrhea and gastrointestinal complaints, in turn improving HRQoL of the patient10.

Conclusion

In conclusion, within the limitations of the current analysis, the economic model indicates that pancreatin MMS treatment of PEI in CP patients is a cost-effective alternative compared to not treating PEI. Pancreatin MMS also remains cost-effective through nearly all of the structural uncertainties tested during the scenario analysis. To our knowledge this is the first cost-effectiveness analysis that assesses the benefit of PERT in CP patients with PEI. Additional long-terms studies on CP-related PEI that provide further data on clinical outcomes, HRQoL, patient statistics and costs (direct and indirect) are recommended in order to perform a more robust analysis in the future.

Transparency

Declaration of funding

The research described in this article, the drafting of this manuscript, and its publication were funded by Abbott Products Operations A.G., Switzerland.

Declaration of financial/other relationships

Anke van Engen, Julia Morawski, and Anja Prüfert have disclosed that they are employees of Quintiles Consulting and that funding for their medical writing support came from Abbott A.G. Douglas Foerster and Suntje Sander-Struckmeier are employees of Abbott. Raffaele Pezzilli contributed as co-author and has collaborated in the past on both Abbott and Quintiles sponsored projects. Ewa Małecka-Panas contributed as a co-author and has also collaborated with Abbott in the past on sponsored projects and events. All authors have participated in the design, data analysis, and drafting of this article.