Melanoma-related costs by disease stage and phase of management in Ireland

Abstract

Background

Management options for the treatment of melanoma have expanded in recent years. In an era of promising, but expensive novel pharmacological treatments, robust stage-specific melanoma-related cost estimates are necessary to support budgetary planning, evaluation of cost-effectiveness and to contribute to the investment case for prevention.

Methods

A detailed decision model, describing the melanoma care pathway (by disease stage) from diagnosis, through treatment and follow-up was developed over a 5-year time frame from the perspective of the Irish healthcare system. The model was populated with real-world data from the National Cancer Registry Ireland. Uncertainty was explored using one-way and probabilistic sensitivity analysis.

Results

The cost of managing a case of melanoma diagnosed at Stage IV (€122 985) was more than 25 times more expensive than managing a case diagnosed at Stage IA (€4269). Total costs were sensitive to the choice of immunotherapeutic and targeted drug, duration of treatment and proportion of patients receiving immunotherapy agents.

Conclusions

The rising incidence of melanoma and high cost of new novel therapies presents an immediate challenge to cancer control and public health globally. This study highlights the cost differential between early and late detection and the potential return on investment for prevention versus high-cost treatment.

Melanoma is the fifth most commonly diagnosed cancer in both men and women in Europe¹ and has one of the fastest growing incidence rates globally.² If 2020 rates continue, the global burden from melanoma is estimated to increase by ~50% by 2040 presenting an important challenge to cancer control and public health globally.²

In Ireland, over 1000 people are diagnosed with melanoma annually, a number which is predicted to triple between 2015 and 2045.³ Harmful exposure to UV radiation (mostly solar radiation) is the most avoidable cause of melanoma risk and mortality, with many economic evaluations reporting favourable benefits from preventative intervention.⁴ Although most cases are caught early, 10% of melanomas in Ireland are still diagnosed at Stage III or IV where 5-year survival rates for distant disease are much lower than for local disease (94.60 versus 32.68%).⁵

Recent advances in treatment have transformed the care of melanoma patients, delivering meaningful improvements in survival; however, the extremely high cost of such treatments presents a challenge for funders.⁶ The impact of early detection on extended survival and subsequent treatment costs has increased the interest in estimating the financial costs associated with melanoma and the associated cost-effectiveness of melanoma screening strategies.^4,7,8

Numerous studies have explored the cost burden of melanoma;^4,9–36 however, many of these findings are now obsolete as they were undertaken before the introduction of high-cost systemic drugs used for adjuvant and advanced melanoma treatment.

The aim of this study was to provide robust estimates of the stage-specific costs of melanoma over a 5-year time frame from the perspective of the Irish healthcare system to inform budgetary planning and underpin future economic evaluation of primary prevention to enable efficient allocation of public funds.

Methods

A model of cumulative cost depending on the initial stage at diagnosis (Stage 0/1A, Stage IB, Stages IIA–IIC, Stages IIIA–IIIC and Stage IV) was constructed, following the newly diagnosed patient for 5 years, when lymph node, in-transit or distant metastases are most likely to occur resulting in diagnostic procedures, surveillance and additional treatment. Tromme et al.⁹ developed a model to estimate the stage-specific economic burden of melanoma in Belgium (Fig. 1), which has subsequently been used to estimate costs in Brazil.³³ We adopted a similar approach, modifying the model to reflect local data availability. In brief, during each annual cycle, and for each stage of disease, patients have a probability of transitioning to another stage. The model incorporated ‘time-dependence’ such that the risk of progression, recurrence or death was dependent on duration in each stage. Our approach differed from Tromme et al.⁹ as we had access to disaggregated data for Stage IIIB/C (which was unavailable from the Belgian registry) hence in-transit metastases were included as Stage IIIB; death was not disaggregated into ‘deaths from melanoma’ and ‘other causes’. In the interests of parsimony, we did not explicitly account for costs associated with false positives in the diagnostic phase of treatment or for multiple lesions. Transition probabilities were assumed the same for both genders and all ages and strictly local recurrences were not included because of their rarity.

Fig. 1.

Melanoma disease model (adapted from that proposed by Tromme et al).⁹

A melanoma care pathway was developed for each stage of disease based on population-level data supplied by the National Cancer Registry Ireland (NCRI), where completeness of registration is estimated to be in excess of 96% of all diagnosed cancers.³⁷ Once developed, the care pathway was reviewed by a multidisciplinary team comprising dermatologists (3), a medical oncologist, a health economist, public health medicine consultants (2), oncology pharmacists (2) and representatives from the National Cancer Control Programme (NCCP) (4) who were convened to oversee the construction, population and validation of the model. Additional expertise was supplied by the NCRI, the Healthcare Pricing Office (HPO) and the Primary Care Reimbursement Service (PCRS) of the Health Service Executive (HSE) in Ireland.

All patients entering the melanoma care pathway were referred by their GP (via the e-referral system) to the skin cancer clinic. Here they were diagnosed, staged (including imaging, dependent on initial preliminary or succeeding clinic-pathological staging), discussed at a melanoma multidisciplinary team meeting (MDTM) and given a definitive diagnosis based on a surgical excision of the melanoma.

Patients with Stage 0 or IA disease underwent no further treatment. A proportion of Stage IB and IIA patients underwent sentinel lymph node biopsy (SLNB) and (where appropriate) were discussed at an Oncology MDTM (on at least one occasion). Patients with a positive SLNB underwent follow-up completion lymphadenectomy. Patients with Stage IV disease underwent surgery or stereotactic radiation for removal or treatment of oligometastatic melanoma. A proportion of patients with Stage IV BRAF mutant melanoma were treated with combination BRAF_MEK inhibitors (it should be noted that the reporting period partly precedes national reimbursement approval of adjuvant Dabrafenib and Trametinib for public patients with Stage III melanoma). Reimbursement approval of these drugs will have financial implications arising from increased prescribing.

A schedule of follow-up visits specific to each disease stage was detailed in the care pathway. Patients with Stage 0/1A disease had two appointments over the following 12 months. The frequency of clinic visits increased with stage of disease (e.g. patients in Stage IIIB/C had four appointments per year in Years 1–3 and twice annually in Years 4–5). In addition, a schedule of investigations accompanied such visits (biannual CT PET + MRI, Years 1–3; annually thereafter until Year 5). Patients suitable for systemic therapy followed a separate schedule of clinic visits, blood tests and imaging.

Data used to populate the model

Data on the distribution of patients across disease stages and resource use estimates (by stage) were supplied by the NCRI for all patients with an ICD-10 principal code for malignant melanoma of the skin (C43). At the time of this study, the most recent available data were for 2016 (n = 1140 patients). The probability of a patient receiving a treatment, alone or in combination, was tabulated by stage and site (Table 1). Additional data (based on ICD-10 principal code) were obtained from the Hospital Inpatient Enquiry team. Supplementary data relating to diagnostic work-up and adherence to surveillance protocols were obtained by undertaking a retrospective chart review of melanoma patients from a large teaching hospital (n = 70, follow-up from 12 to 40 months).

Table 1.

Process and clinical probabilities by stage of disease

Stage of disease	IA	IB	IIA	IIB	IIC	IIIA	IIIB	IIIC	IV
Process probabilities (%)
Excision/destruction skin lesion	100	100	100	100	100	100	100	100	51
Biopsy skin/subcutaneous tissue	14	8	12	17	10	15	6	31	24
Biopsy LN	0	3	1	4	4	18	6	38	24
Biopsy other organ/tissue	0	0	0	1	0	0	0	0	32
Sentinel LN procedure	3	35	32	32	16	60	50	23	3
Regional excision LNs	0	1	0	1	0	33	13	12	3
Radical excision LNs	0	0	0	4	4	38	25	81	14
Other LN excision	0	7	5	9	4	11	13	12	3
Excision lesion other organ/tissue	0	0	0	0	0	0	0	8	19
Amputation fingers/toes	0	0	0	4	2	5	6	0	0
Radiotherapy	0	1	1	1	4	0	19	35	32
Med oncology	0	1	0	0	2	24	13	31	59
Clinical probabilities (interval-specific observed survival) (%)
Year 1	99.09	99.14	97.85	94.85	85.64	100	90.02	82.75	46.54
Year 2	99.14	98.19	94.09	92.2	85.02	96.83	89.69	80.56	79.59
Year 3	98.55	97.91	98.03	94.69	76.56	94.73	92.72	88.03	88.23
Year 4	98.37	98.14	92.96	88.68	76.32	100	91.84	66.67	100
Year 5	99.33	97.08	96.88	95.45	86.67	100	87.49	100	100

In Ireland, data on medication use are captured by the PCRS database; however, it was not possible to isolate resource use by ICD-10 principal code. Data on treatment regimens (by stage) are collected by NCRI; however, this does not extend to the dose and duration of radiotherapy, chemotherapy, immunotherapy and targeted drugs. Hence conservative estimates based on NCCP approved treatment regimens for melanoma and validated by expert opinion within the group (Oncology Consultant and Senior NCCP pharmacists) were employed.

Following treatment, individuals have a stage-specific risk of remaining in remission, having a recurrence, distant metastases or death. In the absence of data on disease free progression for this patient population, data on the probability of dying in a 12-month period (using interval-specific observed survival by stage for Years 1–5) were supplied by the NCRI (Table 1). The probability of recurrence was based on recurrence-free survival rates presented in Leiter et al.³⁸ and further assumed that 77% of initial recurrences were locoregional with the remainder classified as distant metastases.

Valuation of resources

Direct healthcare costs were collected from the perspective of the publicly funded Irish health and social care system (HSE) for 2020 (€). Costs associated with diagnosis, in-patient and out-patient procedures, systemic therapy, relapse and supportive care were included. All costs were adjusted to 2020 (€) using the Consumer Price Index for Health (Central Statistics Office www.cso.ie) and discounted at a rate of 4% per annum as per best practice guidelines for performing Health Technology Assessment in Ireland.³⁹

Nationally produced unit costs for individual procedures are not produced in Ireland, hence procedures captured by NCRI were mapped across to Diagnosis Related Groups codes (DRGs) compiled by the Health Pricing Office (HPO) using the ICD-10 Principal code for Melanoma (C43). This work package involved assessing the frequency with which procedures were performed in conjunction with expert opinion from HPO senior analysts and clinical expert opinion to construct frequency weighted mean costs (reflecting level of complexity and whether it was performed as an inpatient or day case procedure). Analysts in HPO were consulted to ensure the ‘DRG cost bundle’ avoided double counting (specifically with respect to blood tests and imaging). Unit costs are presented in Table 2; diagnosis, treatment and surveillance costs are presented for Years 1–5 (by stage) in Supplementary Table S1.

Table 2.

Unit costs for procedures (ABF Admitted Patient Price List, €2020)

DRG code	Medical or surgical	Definition	Day case prices (2020, in €)	Inpatient prices (2020, in €)
J08A	Surgical	Other Skin Grafts and Debridement Procedures, major complexity	5394	20 423
J08B	Surgical	Other Skin Grafts and Debridement Procedures, intermediate complexity	3903	7921
J08C	Surgical	Other Skin Grafts and Debridement Procedures, minor complexity	1844	5510
J10A	Surgical	Plastic GIs for Skin, Subcutaneous Tissue and Breast Disorders, major complexity	3410	6907
J10B	Surgical	Plastic GIs for Skin, Subcutaneous Tissue and Breast Disorders, minor complexity	1935	4964
J11A	Surgical	Other Skin, Subcutaneous Tissue and Breast Procedures, major complexity	1283	6135
J11B	Surgical	Other Skin, Subcutaneous Tissue and Breast Procedures, minor complexity	678	3459
J13A	Surgical	Lower Limb Procedures W/O Ulcer or Cellulitis, major complexity	3993	10 405
J13B	Surgical	Lower Limb Procedures W/O Ulcer or Cellulitis, minor complexity	2509	4904
J69A	Medical	Skin malignancy major complexity	1292	13 089
J69B	Medical	Skin malignancy intermediate complexity	853	7168
J69C	Medical	Skin malignancy minor complexity	494	6257
R62C	Medical	Other neoplastic disorders, minor complexity (radio)	282	2933
R63Z	Medical	Chemotherapy	557	—
R99Z	Medical	Oncology repeat attendance	525	—
G70B	Medical	Other digestive system disorders	694	—
Q61	Medical	Red blood cell disorders (weighted average of Q61A, Q61B, Q61C)	460	—
R62	Medical	Heart failure and shock (weighted average of R62A, R62B, R62C)	1528	—
F67	Medical	Hypertension (weighted average of F67A, F67B)	352	—
C03A	Surgical	Retinal procedures, major complexity	—	3004
E62B	Medical	Respiratory infections inflammation, minor complexity	—	3509
G64B	Medical	Inflammatory bowel disease, minor complexity	—	3209
K64B	Medical	Endocrine disorders, minor complexity	—	2722

The cost of immunotherapy regimens approved for use by the HSE for treatment of melanoma was provided by the NCCP https://www.hse.ie/eng/services/list/5/cancer/profinfo/chemoprotocols/melanoma/. Costs were based on list price including VAT (23%) plus an administration cost (DRG code R63Z from the Admitted Patient Price List 2020 for chemotherapy day case) (Supplementary Table S2). The monitoring and management of adverse events (AEs) (which is not included in the administration cost DRG) were calculated separately based on reported AE frequency from clinical trials and using a list of costs from the ABF Admitted Patient Price List 2020 (Supplementary Table S3; https://www.hpo.ie/abf/ABF2020AdmittedPatientPriceList.pdf). In the base-case analysis, costs associated with 12 months of therapy were applied to all patients residing in that state.

The cost of supportive care was derived from estimates of the cost of palliative care in the Irish healthcare system utilizing only that portion of cost attributable to specialist palliative and hospital care costs (€11 685).⁴⁰

Sensitivity analysis

One-way and probabilistic sensitivity analyses (PSA) were undertaken to explore the impact of parameter uncertainty on the total cost estimates. Parameters associated with most uncertainty were subjected to one-way sensitivity analysis. PSA were conducted by running the model with a hypothetical cohort of 10 000 patients calculating cost estimates for each simulation. In each simulation, the value for each parameter was varied across a probability distribution reflective of the properties of the parameter under consideration and data informing them. Probabilities were modelled using beta distributions and parameterized with the sample size from the relevant published literature or data from NCRI. Costs were specified as gamma distributions as they are constrained to be positive, continuous and highly skewed.

All analyses were undertaken in TreeAge Pro 2021 (TreeAge Software, Inc., Williamstown, MA) and reported in line with The Consolidated Health Economic Evaluation Reporting Standards 2022 statement.⁴¹

Results

The estimated costs per person diagnosed with melanoma over a 5-year period are set out in Table 3. Patients with the best prognosis (Stage IA) were the least costly to treat and manage compared with those with the worst prognosis (Stage IV). The cost of managing a case of melanoma diagnosed at Stage IV was over 25 times more expensive than managing a case diagnosed at Stage IA (Stage IV €122 985 versus Stage IA €4269).

Table 3.

Summary of costs of managing melanoma by stage of disease (€2020), base-case estimates and selected sensitivity analyses (95% CI from PSA)

Stage at diagnosis	Base-case estimates (€)	95% CI from PSA (€)
IA	4269	3594–4984
IB	6313	5444–7275
IIA	15 130	12 301–18 370
IIB	15 291	12 571–18 330
IIC	14 905	12 711–17 324
IIIA	92 762	61 144–138 637
IIIB	102 951	68 480–145 727
IIIC	113 655	80 802–150 356
IV	122 985	84 044–167 606
Stage weighteda	20 268	16 126–25 071
	Sensitivity analysis (change in proportion of patients on systemic therapy)
IA	5188	4501–5637
IB	7229	6828–7672
IIA	22 478	19 258–25 921
IIB	22 162	19 232–25 301
IIC	20 223	17 959–22 653
IIIA	223 769	171 818–273 194
IIIB	239 667	201 867–278 490
IIIC	229 868	195 701–266 312
IV	230 573	186 863–277 182
Stage weighteda	37 715	32 619–43 019
	Sensitivity analysis (change in recurrence rate)
IA (20% lower recurrence rate)	4067	3429–4750
IA (20% higher recurrence rate)	4733	3987–5528
IIA (20% lower recurrence rate)	13 015	10 817–15 509
IIA (20% higher recurrence rate)	17 346	13 898–21 264
IIIA (20% lower recurrence rate)	84 911	55 028–130 332
IIIA (20% higher recurrence rate)	100 025	66 641–146 814
Stage weighteda (20% lower recurrence rate)	19 217	15 388–23 680
Stage weighteda (20% higher recurrence rate)	21 476	13 365–26 658

^aStage weighting was based on NCRI data (IA 44%, IB 25%, IIA 7%, IIB 6%, IIC 5%, IIIA 5%, IIIB 1%, IIIC 2% and IV 4%).

As can be seen costs escalate in part because of increased activity—for example SNLB in Stage II—and in part because of the deployment of more expensive therapies associated with disease progression.

Sensitivity analysis

The 95% confidence interval (CI) for the average (stage-weighted) cost per person for managing a case of melanoma over a 5-year period (base case €20 268) was estimated from the PSA and ranged from €16 126 to €25 071 (Table 3).

To reflect changes in practice since data were obtained from NCRI (for 2016), we assumed that 70% of patients in Stage IIIA and 100% of patients in Stages IIIB and IIIC received immunotherapy or targeted drugs for a maximum of 12 months, and 100% of Stage IV patients for a maximum of 24 months. This assumption was based on expert opinion within the group. Under these assumptions the stage-weighted average cost per person almost doubled from €20 268 (CI: €16 126–€25 071) (Table 3) to €37 715 (CI: €32 619–€43 019) (Table 3). In the base-case analysis, we assumed that all patients consumed 100% of prescribed medication. When we assumed that 88% of doses were consumed (as per Pike et al.²⁹) the stage-weighted average cost per person fell to €18 561 (€14 915–€22 793). A flat-rate of 88% was used; however, it is noteworthy that Pike et al.²⁹ also adjusted the quantity of doses consumed downward on a monthly basis by drug type.

In the base-case analysis, we assumed that 77% of recurrences were locoregional, and the rest were distant. When we assumed a 20% reduction in locoregional progression, the stage-weighted average cost per person rose to €21 079 (€16 681–€26 180) and similarly fell to €19 461 (€15 569—€23 988) when we assumed a 20% increase.

Interval-specific recurrence rates for each stage were employed in the model.³⁸ When these recurrence rates were varied by 20% around the base case estimate, the stage-weighted average cost per patient changed from €19 217 (CI: €15 388–€23 680) to €21 476 (€13 365–€26 658). Figures for selected stages are presented in Table 3.

Discussion

Main findings of this study

Healthcare costs for those diagnosed with Stage IV disease (over a 5-year time horizon) were over 25 times higher than for those diagnosed at Stage IA. Given prognosis is also superior, it underscores the importance of early detection and intervention. The cost of managing melanoma was particularly sensitive to those factors that contribute to the cost of immunotherapeutic and targeted drugs (such as dose, duration and choice of agent). Despite the use of sensitivity analysis to explore potential sources of uncertainty, it is clear that the collection of ‘real-world’ data relating to ‘key’ cost drivers is crucial if we are to understand the distribution of costs and accurately estimate the cost-effectiveness of early intervention and treatment strategies.

What is already known on this topic

Our study echoes findings in the literature regarding the relationship between disease stage and cost.^{10,20–22,27,32,33} However, the difference in costs between stages has changed dramatically with the introduction of immunotherapeutic and targeted drugs rendering findings from earlier papers less relevant.

Our findings support those found in other recent studies, which have differentiated cost by stage of disease. One-year costs within the Italian health service ranged from €1 837 to €66 950 producing a cost ratio of 31.27 (between Stages I and IV).³² In the Spanish health care system, costs estimated over a 10-year time frame ranged from €1689 (Stage I) to €88 268 (Stage IV) reporting a doubling in costs where there was lymph node involvement, and treatment costs that were 52-fold higher for those with the least as opposed to most favourable prognosis.³¹ Similarly, spending on advanced melanoma in Brazil (estimated over a 3-year time frame) was reported as being up to 34-fold higher in the public health system compared with lower-stage disease.³³ In the Netherlands, the real-world healthcare costs of melanoma were explored, and since the introduction of immunotherapeutic and targeted drugs, systemic therapies now account for 84% of costs, with costs differing substantially between patients who receive systemic therapy (€105 078) and those who do not (€7988).⁶

What this study adds

This is the first study to estimate the healthcare costs for melanoma patients by disease stage in Ireland. A strength of this study is that real-world data were used to populate the model that contrasts with other studies that have estimated resource utilization using national guidelines, ‘best practice’ or expert opinion/elicitation. Our study will enable robust estimation of the cost-effectiveness of prevention strategies in the Irish healthcare setting, and provide a blue-print for estimating the stage-specific costs of other cancers in Ireland. It has also clarified where best to focus resources with respect to data collection and linkage efforts moving forward.

Limitations of this study

The study has a number of limitations: this study used data preceding the adoption of the American Joint Committee version 8 (which includes Stage IIID), which may affect a small number of transitions; we did not explicitly account for costs associated with false positives in the diagnostic phase of treatment or for multiple lesions; however, the marginal cost of both is likely to be small; recurrence data from the literature was used to inform disease progression which underestimated 5-year survival in the Irish context (potentially underestimating costs); we relied on published NCCP regimens for dose and duration information and expert opinion relating to medication use as these data were unavailable; we did not vary recurrence rates or survival by age and/or gender; and patients with Stage IV disease were not segregated by site of recurrence, despite this potentially impacting both survival and costs.

Currently over 10% of melanomas are diagnosed in Ireland at Stage III or IV (NCRI). Evidence regarding the potential savings associated with earlier identification and treatment are vital to equip those interested in provoking a policy response in that direction. Lower costs and superior outcomes associated with early identification and treatment are especially important given current trends with respect to incidence and the cost of novel therapies. Greater awareness and increased capacity for diagnosis in the publicly funded system have the potential to improve system efficiency and equity; therefore, this topic must remain high on the public health agenda in Ireland.

Grainne E. Crealey, Health Economist

Caitriona Hackett, Consultant Dermatologist

Katharine Harkin, Specialist in Public Health Medicine

Patricia Heckmann, Assistant National Director

Fergal Kelleher, Consultant Medical Oncologist

Áine Lyng, Cancer Prevention Officer

Triona McCarthy, Consultant in Public Health Medicine

Maria McEnery, Cancer Prevention Officer

Clare Meaney, Senior II Pharmacist

Darren Roche, Dermatology SPR

Anne-Marie Tobin, Consultant Dermatologist

Funding

The National Cancer Control Programme (NCCP) commissioned this work with G.C. as an external consultant in health economics working with the assembled team.

Data availability

The data underlying this article are available in the article and in the supplementary tables supplied.

Conflict of interest statement

None of the authors have a conflict of interest to declare.