In the USA, skin cancer is a major public health problem with increasing incidence and annual treatment costs estimated at $8.1 billion . Melanoma, the most serious type of skin cancer, is listed as the fifth most common cancer in the USA. Once it metastasizes, it has historically been considered a treatment-resistant malignancy.
The armamentarium of melanoma treatments, previously limited to traditional surgery, radiation therapy, and chemotherapy, has now expanded to include immunotherapy and targeted therapy. Until 2011, dacarbazine and high-dose interleukin 2 were the only agents for metastatic melanoma approved by the US Food and Drug Administration. Presently, there are four approved immunotherapy agents (ipilimumab, pembrolizumab, nivolumab, and interleukin 2) and four approved targeted therapy agents (vemurafenib, dabrafenib, trametinib, cobimetinib) approved by the Food and Drug Administration [2–7]. A recent publication reported a 3-year overall survival (OS) rate of 58% for combination nivolumab and ipilimumab in patients with advanced melanoma . New agents afford patients with advanced melanoma more treatment choices.
Corollary to this expansion is an increase in the complexity of healthcare decision-making. The Institute of Medicine defines high-quality healthcare as good communication, shared decision-making, and respect for patients’ values and preferences . Although value assessment of therapy has traditionally focused on survival advantage and economic gain, the real-world definition of value may differ by what is important to a particular audience (patient, provider, payer, policy maker, etc.) . For example, among patients with solid tumors, 77% indicate a preference for ‘hopeful gambles’ in treatment selection, where a low percentage of patients derive long-term survival, compared with ‘safe bet’ treatments conferring a higher percentage of patients with an average survival benefit but no chance at a large gain .
As the molecular underpinnings of melanoma continue to be explored, new therapies will become available. Thus, it is important to understand how patients value therapy and to elucidate what facets of treatment are important to them. A personalized treatment strategy that incorporates patients’ perceptions about melanoma, prognosis, treatment characteristics, expected clinical outcomes, and overall desired quality-of-life may contribute to treatment acceptance, utilization, and adherence. To address these gaps in patient and provider preferences surrounding melanoma treatment, we performed a prospective survey study using a discrete choice experiment (DCE) design that was administered to both patients with melanoma and their corresponding providers. In addition to characterizing patient and provider preferences, we calculated the respective willingness-to-pay (WTP) for immunotherapy and targeted therapy for melanoma treatment.
Patients and methods
A three-part tailored survey was built separately for patients and providers (see ‘Patient and Provider Surveys’ section, Supplemental digital content 1, http://links.lww.com/MR/A106). The first part of the survey assessed patient and provider demographics, disease and treatment characteristics (patients only), and perceptions and concerns toward melanoma care, treatment, and costs. The second part of the survey assessed the willingness of patients and providers to receive and/or recommend different systemic melanoma treatments, respectively, based on differing effectiveness and safety parameters. The last portion of the survey was the DCE. The questionnaire was developed at a fifth grade literacy level, and the contents were validated by provider focus groups and patient interviews. In addition, patient-friendly graphics were included to help illustrate/visualize the patient’s experience.
Discrete choice experiment
DCE is a rigorous method for examining preference and WTP in healthcare . It includes various choice sets, with each one consisting of different hypothetical treatment alternatives. These alternatives differ by treatment attributes, such as efficacy, side effects, cost, and levels of attributes. For each choice set, alternatives are compared and one alternative is selected. Based on random utility theory, a multinomial logit model was developed to determine the influences of attributes on patient preference and WTP for each level of change across all attributes.
Attribution and level identification
A literature review of studies reporting efficacy and safety of immunotherapy and targeted therapy in metastatic melanoma was conducted to determine treatment attributes. The initial list of attributes included: 5-year OS, 3-year OS, median OS, median progression-free survival, time to response, duration of response, immunotherapy-related side effects, skin toxicity, and gastrointestinal (GI) side effects. This list of attributes was discussed with providers and patients (details below). Eventually, six attributes including median OS, immunotherapy-related side effects, skin toxicity, GI toxicity, route of administration, and monthly out-of-pocket cost (OOP) were selected for the study (Table 1). For each attribute included in the DCE, the minimum and maximum values were obtained by literature review (Supplementary Table 1, Supplemental digital content 1, http://links.lww.com/MR/A106). Each attribute had three levels, except the route of administration and cost attributes, which had two and six levels, respectively. The route of administration included oral and intravenous dosage forms given the type of treatment availability for melanoma. The ranges for each treatment attribute were divided equally per level. The cost attribute was defined as the patient’s monthly OOP cost derived from an estimated annual systemic treatment cost of $150 000 with an 80% insurance coverage rate. OOP costs were defined to patients and providers as the monthly amount for medications they would be responsible for paying OOP above and beyond their medical or prescription drug insurance benefits. It has been reported that 15 and 19% of cancer patients indicate spending from $2500 to less than $5000 to treat their cancer, including medications, in the past 12 months and since initial diagnosis, respectively .
As it was not feasible to present all possible combinations of selected attributes and levels (3×3×3×3×2×6) to each individual patient or provider, Ngene software (version 1.1.1, Choice-Metrics, Sydney, Australia) was used to draw a subset of all combinations using an orthogonal and level balance design. Two different 36-choice sets were generated and divided into six and two blocks for patients and providers, respectively. Therefore, there were six and two different versions of the questionnaires for patients and providers, respectively. Each patient questionnaire contained six different choice sets, and each provider questionnaire contained 18 different choice sets. Each choice set consisted of two hypothetical melanoma treatments and an opt-out alternative to allow patients and providers to choose neither treatment. Pictograms were added to help patients understand the attributes, particularly the adverse events associated with the treatments. Table 1 shows an example of the choice sets. Each questionnaire also contained an additional choice that served as a validity check. This choice set contained a dominant alternative (highest OS, lowest immunotherapy-related side effect, skin toxicity, GI toxicity, oral formulation, and lowest OOP cost) that patients or providers who understood the DCE questions must choose.
Focus groups, patient interviews, and pilot study
Two focus groups were conducted with melanoma specialists and three interviews with patients with melanoma. In total, two medical oncologists and two surgical oncologists provided important aspects of melanoma care from the provider perspective. Three individual patients were interviewed, who were receiving care from the specialists and who collectively represented the spectrum of melanoma disease and treatment exposure. During the focus groups and interviews, providers and patients were asked to review the initial list of attributes of melanoma treatments. They were also asked to rank the importance of the efficacy and safety attributes of melanoma treatment to gain consensus in determining the most important attributes to include in the survey. After the survey was developed, it was fielded in a pilot phase. A convenient sample of 10 patients from the eligible cohort, not including the three patients previously interviewed, evaluated the survey’s layout, comprehensibility, graphics, content relevance, and questions.
The cohort of eligible study participants with melanoma was created using the University of Utah Health Sciences Enterprise Data Warehouse and the Huntsman Cancer Institute Tumor registry (HCI-TR). Inclusion criteria included the following: patients at least 18 years at the time of diagnosis; at least 2 ICD-9 codes for melanoma; inclusion in the HCI-TR with International Classification of Diseases for Oncology, 3rd ed., site and histology codes indicative of melanoma with stage I–IV disease between 2013 and 2014; and at least two encounters in the Enterprise Data Warehouse, with one encounter at least 30 days from the index date; a valid e-mail address on record; and patients were required to be alive to actively participate in the survey. Eligible providers included physicians, pharmacists, and nurses at HCI involved in providing direct care to patients with melanoma or whose clinical activities involved immunotherapy or targeted therapy. Institutional Review Board approval was granted from the University of Utah.
The final self-administered questionnaire was built using Qualtrics software (Provo, Utah, USA). Survey dissemination spanned January 2016 to March 2016. Institutional Review Board approved recruitment and survey materials were initially distributed by e-mail to eligible patients and providers. The introductory e-mail included a survey link and a study telephone number, which was followed up with a reminder phone call or e-mail ~2 weeks later. Phone calls were placed to patients to ensure receipt of the e-mail invitation and to answer any questions. Patients who remained interested in the study could complete the survey electronically or were offered a duplicate paper version of the survey mailed with a prestamped return envelope. Instead of follow-up phone calls to providers, reminder e-mails were sent and paper versions of the survey were delivered to their offices. All study participants were offered a nominal monetary incentive or a donation to a cancer organization on their behalf as a token of appreciation. Participants were given 3 months to complete the survey. Surveys completed through the e-mail link were automatically captured. Responses provided on the paper version were manually entered into Qualtrics.
Survey responses for the demographics and general survey sections were descriptively analyzed using χ2-test or Fisher’s exact test as appropriate. Demographics were compared between patient responders and nonresponders using data contained in the Enterprise Data Warehouse to determine underlying differences.
According to good DCE research practice, the aim was for 150 complete and valid responses from at least 150 patient participants to ensure study precision . For the DCE section, a multinomial logit model based on random utility theory was developed to estimate the following utility model: Uij, that either a patient or a provider i assigned to an alternative I, from patients’ and providers’ responses for choice sets using NLOGIT version 4.
where β0 is the constant reflecting patients’ or providers’ preference for selecting melanoma treatment relative to no treatment, and β1, β2, β3, β4, β5, and β6 are the coefficients or the mean attribute weights of OS, immunotherapy-related side effect (IM), skin toxicity (SK), GI, route of administration (AD), and monthly OOP cost (cost), respectively, and εij is error term. The magnitude of each coefficient indicated the relative importance of each attribute, whereas the sign of the coefficient reflected whether the attribute had a positive or a negative effect on utility or preference, as compared with the base level. The level of statistical significance was set at 0.05.
For each attribute, the marginal WTP, which indicates how much a patient or a provider is willing to pay for a one-unit change in the attribute, was calculated by taking the ratio of the mean attribute coefficient to the mean coefficient of cost attribute. The Krinsky and Robb  method was used to estimate the 95% confidence interval of the WTP for each attribute. Finally, the WTP for each existing melanoma treatment was calculated by multiplying the marginal WTP for all attributes and their level changes, which were obtained from literature (Supplementary Table 1, Supplemental digital content 1, http://links.lww.com/MR/A106). Simple one-way sensitivity analysis was performed, varying attribute levels from the lowest to highest point estimates.
Patient and provider characteristics and their perceptions of melanoma treatments
An eligible cohort of 557 patients was initially identified for the study from 2013 to 2014. A total of 233 patients completed the survey (electronic version, n = 91; paper version, n = 142) reflecting a 42% response rate. Fifty-three providers were invited to participate in the study, with 20 providers completing the survey (electronic version, n = 15; paper version, n = 5) for a 38% response rate.
Table 2 presents the demographics and disease characteristics of the patient respondents, demonstrating no difference compared to nonrespondents (Supplementary Table 2, Supplemental digital content 1, http://links.lww.com/MR/A106). The median age was 64 years and 62% were male. The majority of respondents were well-educated (42% college education), had insurance to cover their melanoma treatment (94%), and their annual median OOP cost for medical treatment was $2500 (mean $4621). A majority of respondents were diagnosed with early-stage melanoma, stage I or II (75%), and most were not currently receiving active treatment for their disease (85%). Prior treatments for melanoma were reported as surgery (92%), systemic treatment (12%), and radiation (7%). Immunotherapy was administered to 18 (8%) patients, 12 (5%) patients had received chemotherapy, and BRAF/MEK inhibitors were administered to four (2%) patients.
Table 3 presents the provider respondents’ characteristics. The median age of providers was 38.5 years and 40% were male. Seven oncologists, nine oncology pharmacists, and four nurses responded to the survey. The average number of patients treated per year was 390 patients and the average number of years treating cancer patients was 7.3 years.
Figure 1 presents patient and provider quality-of-life expectations, relations, and perceptions toward melanoma and its treatment. More providers than patients selected ‘feeling less pain’ as a quality-of-life indicator (65.0 vs. 28.6%; P < 0.001; Fig. 1a). Most providers perceived patients were ‘very anxious’ about their melanoma, whereas most patients reported feeling ‘not anxious’ (P < 0.0001; Fig. 1b). Almost half of patients responded that they ‘always’ have enough time to discuss their melanoma treatment with their provider, while only one provider felt similarly (P < 0.001; Fig. 1c). Approximately 70% of patients indicated ‘always’ trusting their providers to make the best treatment decision compared with 15% of providers that felt they ‘always’ trusted themselves (P < 0.0001; Fig. 1d). Additionally, most patients (74%) reported ‘always’ feeling that their providers share their opinions about melanoma, compared with most providers (39%) selecting ‘rarely’ (P < 0.0001; Fig. 1e). Conversely, patients reported ‘always’ or ‘most of the time’ sharing their concerns, whereas most providers felt this was ‘sometimes’ true (P < 0.0001; Fig. 1f). Patients most frequently responded that providers ‘sometimes’ understand the impact of their OOP costs for melanoma treatment, whereas providers most frequently responded ‘most of the time’ (P = 0.002; Fig. 1g). Sixty percent of providers would recommend melanoma treatment that would be effective for 6–11 months. Approximately 50% of patients would only elect to undergo treatment if it was effective long-term, specifically for more than 24 months (P = 0.007; Fig. 1h).
Patients’ and providers’ preferences and their willingness to pay
Table 4 shows the estimated parameters from the multinomial logit models. No bad observations (failing validity check) were included in the DCE analyses for both patient and provider groups. Approximately 29% of patients and 8% of providers chose the no treatment alternative, respectively, meaning they would not choose either treatment. For each group, the remainder of the selected treatment alternatives were distributed similarly between the two treatment alternatives. Among the six tested attributes, only OS, immunotherapy-related side effect, skin toxicity, and monthly OOP cost were statistically significant and had expected signs in both models. The positive sign of the estimated coefficients for OS reflects that both patients and providers preferred melanoma treatments that extended survival. Conversely, the negative signs of the estimated coefficients for immunotherapy-related side effects, skin toxicity, and monthly OOP cost indicated respondents prefer less frequent side effects and toxicity, and lower monthly OOP cost. Among all four significant attributes, only the coefficient magnitudes of the immunotherapy-related side effect and skin toxicity can be compared directly because their levels had the same unit (frequency of side effect). The patients weighted the immunotherapy-related side effect almost three times the weight of skin toxicity (0.033 vs. 0.013), whereas the provider evenly weighted these attributes (0.075 vs. 0.070).
Table 5 shows the patients’ and providers’ marginal WTP per month per unit change in the attribute’s level. The patients and providers were willing to pay $932 and $1008 per month, respectively, for every 1-year increase in OS. The patients and providers were willing to pay $50 and $55 per month, respectively, to reduce the occurrence of immunotherapy-related side effects by 1%. The providers were willing to pay $31 more per month than patients to reduce the occurrence of skin toxicities by 1% ($51 vs. $20). Finally, the attribute levels of combination immunotherapy (nivolumab + ipilimumab) and BRAF/MEK inhibitors extracted from the literature (Supplementary Table 1, Supplemental digital content 1, http://links.lww.com/MR/A106) were used to calculate the WTP for these treatments. Patients and providers were willing to pay $2357 and $2484 per month OOP, respectively, for combination immunotherapy. For BRAF/MEK inhibitors, patients and providers were willing to pay $1648 and $1350 per month OOP, respectively. Based on the 95% confidence intervals for survival estimates for immunotherapy and BRAF/MEK inhibitors, one-way sensitivity analysis determined patient’s WTP for immunotherapy ranged from $1700 to $3020 for an OS distribution of 2.0–3.7 years. For BRAF/MEK inhibitors it ranged from $1072 to $2547 for an OS distribution of 1.5–3.0 years. The sensitivity analysis for the adverse event attributes are presented in the Supplementary Fig. (Supplemental digital content 1, http://links.lww.com/MR/A106).
Shared decision-making between patient and provider is integral to providing optimal cancer care. Previous findings demonstrate a discord between perceived and actual patient requests. Physicians often feel ‘patients who demand interventions or treatments’ promote utilization of costly treatments, despite evidence to the contrary . Other studies have demonstrated that patients prefer treatment that affords a ‘hopeful gamble’ for a small chance to obtain large gains in health outcomes. This contrasts with providers who tend to prefer ‘safer bets’ for small, incremental benefits in health outcomes with relatively high chance of occurrence [9,15]. In our study, exploration of these patient and provider relationships showed a similar discordance, where ~70% of patients ‘always’ trust their providers to make the best treatment decision for them compared with 15% of providers responses who feel that they actually do. This discordance in perception between patients and providers may likely contribute to differing beliefs toward therapeutic decision-making. Contributing factors to these differences, which were demonstrated in our study, include perceptions of patient anxiety toward melanoma treatment, conflicting perspectives on adequate time to discuss melanoma treatment options, and ultimately, optimism for outcomes of various treatment options.
Among the observed attributes in the DCE, OS, immunotherapy-related side effects, skin toxicity, and monthly OOP cost were statistically significant. As expected, our study demonstrated both patient and provider preference for survival benefit in the DCE, which is commonly reported in DCEs in oncology [16–18]. OS was included as an attribute in the DCE due to the profound change in long-term survival with the introduction of these novel treatment options. In clinical trial setting, patients with advanced melanoma have demonstrated a 3-year OS rate of 58% for combination nivolumab and ipilimumab , 44% for combined BRAF/MEK inhibitors , and 21% with ipilimumab + dacarbazine, or 12% with dacarbazine alone . Patients preferred to avoid immunotherapy-related side effects almost three times more than skin toxicity (0.033 vs. 0.013), whereas providers evenly weighted these treatment attributes (0.075 vs. 0.070). Route of administration and mild side effects did not carry as much weight in both patients’ and providers’ decision-making.
WTP was predominantly driven by gains in survival. These data are again similar to prior reports demonstrating the value patients place on survival time [9,15,21]. However, our study was able to explore the subtleties around survival beyond the DCE. Sixty percent of providers would recommend melanoma treatment that would be effective for 6–11 months compared with ~50% of patients who would only elect to undergo treatment if it was effective for more than 24 months. This finding highlights the need for thoughtful discussions with patients about treatment preferences when recommending therapy. Despite the differences in survival expectations for treatment, patients and providers had similar OOP WTP thresholds for OS of ~$900–$1000 per month per 1-year increase in survival. Patients and providers had similar WTP for immunotherapy ($2357 and $2484, respectively). In contrast, patients and providers were willing to pay $1648 and $1350, respectively for BRAF/MEK inhibitors, stemming from the difference in perceived OS benefit between these treatment modalities. Interestingly, providers’ WTP for BRAF/MEK inhibitors was nearly half that of immunotherapy, which may be explained by providers’ knowledge of high resistance rates with BRAF/MEK inhibitors compared with the durable responses that have been observed with immunotherapy [2,22,23]. Overall these OOP WTP costs represent significant portions of the patients’ reported monthly income considering 36% of respondents reported an annual household income of less than $49 999. In addition, both patients and providers were willing to pay, or have their patient’s pay, approximately the same amount to avoid immunotherapy-related side effects suggesting that they viewed immunotherapy-related side effects similarly. Regarding skin toxicities, the providers’ WTP was more than that of the patients. Providers value avoiding skin toxicities more than patients, a preference likely rooted in their experience with managing skin toxicities resulting from melanoma treatments, especially BRAF/MEK inhibitors.
Our study was unique in demonstrating a more granular understanding of the important attributes of treatment options for melanoma compared with other DCEs [16,17]. Unlike previous studies, our study also incorporated a spectrum of oncology-specialized clinicians, including oncology-trained pharmacists and nurse practitioners in addition to the more commonly investigated physicians. This diversification adds greater context to the findings given that pharmacists have intimate knowledge of therapies and nurse practitioners have vast patient interactions in clinical applications. Investigating this wider spectrum of oncology clinicians brings greater context to decision-making in regards to adverse events, common trends in patient preference, and procurement of therapies. However, including a broad spectrum on oncology practitioners may contribute to greater heterogeneity in the expectations of patient outcomes influencing the outcomes of the study.
Three limitations of the study warrant mention. First, our study was performed at an academic cancer center in the Intermountain West staffed by melanoma physicians who are specialists in surgery and oncology. Therefore, the results may not be generalizable to community cancer centers or regions with differing ethnicities. Homogeny in a sample population is often preferable for comparing results in a variety of study types . Our study population included primarily European American, well-educated, insurance-carrying patients with early-stage melanoma. Patients’ overall lack of exposure to systemic treatment would likely minimize bias from previous treatment; however, patients were required to weigh attributes of treatment reserved for advanced melanoma with which they may have little or no actual experience. For example, patients’ WTP to avoid skin toxicities was less compared with providers’ WTP and may reflect lack of personal familiarity with this particular adverse effect. In addition, our study did not control for differences in the cultural background of patients and clinicians or annual incomes levels which could have influenced outcomes and WTP . More heterogeneity in the sample population would have also increased generalizability to a broader population. Second, our one-time survey dissemination represented a snapshot in time of the patient experience with immunotherapy and targeted therapy. The comparative unfamiliarity with systemic treatments and their associated toxicity profile among patients with early-stage melanoma may be addressed in the future with a larger study of patients with advanced disease and treatment experience. Given the more durable response of systemic therapies now reported in recent publications, it is possible to have a larger study comparing the preferences and WTP of patients with early-stage disease with those with advanced disease. Last, a limitation to all DCEs is not being able to investigate all attributes related to a treatment decision. Other attributes such as additional side effects, progression-free survival, and lower-grade adverse events were not assessed in our study, as these were eliminated by the patient or clinician focus groups. In addition, an emerging attribute of combination immunotherapy, treatment-free survival (i.e. being able to maintain durable benefit after stopping treatment), was not assessed and would warrant future study on the value patients and providers place on it.
There was discordance between the patient-provider relationship and perceptions around melanoma treatment. Patients understandably want treatment with longer survival advantage and remain more optimistic, whereas providers tend to make decisions based on reported survival benefit and personal experience using these therapies. Long-term durable benefit is very important to patients, with 50% indicating that they would only elect to undergo treatment if it was effective for more than 24 months. Patient-provider preferences may enrich shared decision-making, especially as more efficacious treatments become available and there is a need to balance survival benefits with side effects and costs. As expected, patients and providers were willing to accept all risk levels from adverse effects of immunotherapy and BRAF/MEK inhibitors for improved survival. Patients and providers exhibited a greater WTP for combination immunotherapy as compared to BRAF/MEK inhibitor therapy, indicating a greater perceived survival benefit with combination immunotherapy.
The authors thank all the focus group and survey participants for their time and insight into the study. We would also like to thank Amy Stanford, Manan Shah, and Tony Okoro for their intellectual contributions to early portions of this research.
This work was supported by Bristol-Myers Squibb.
Project conception/design was done by Diana I. Brixner, David D. Stenehjem, Beata Korytowsky, Gary Oderda. Data collection was done by Junjie Ma, Ryan S. Nelson, Hillevi Bauer, Trang H. Au, Amy Stanford. Data analysis was carried out by David D. Stenehjem, Trang H. Au, Surachat Ngorsuraches, Junjie Ma, Tanatape Wanishayakorn, Amy Stanford. Data interpretation was done by David D. Stenehjem, Diana I. Brixner, Gary Oderda, Trang H. Au, Joshua Schwartz, Manan Shah, Tony Okoro, Surachat Ngorsuraches, Beata Korytowsky. Manuscript writing was carried out by David D. Stenehjem, Trang H. Au, Surachat Ngorsuraches, Diana I. Brixner, Ryan S. Nelson. Project supervision was carried out by Diana I. Brixner, David D. Stenehjem, Gary Oderda, Beata Korytowsky. Manuscript review was done by all authors.
Conflicts of interest
David D. Stenehjem reports relevant research funding from Bristol-Myers Squibb during the course of the study; personal consulting fees from Bristol-Myers Squibb, Salarius Pharmaceuticals, and Iterion Therapeutics outside the submitted work. Dr Diana I. Brixner reports a relevant research grant from BMS, during the conduct of the study; personal fees from AbbVie, Becton Dickinson, Abbott, and Millcreek Outcomes Group, outside the submitted work. Joshua Schwartz, Constance M. Pfeiffer, and Beata Korytowsky are employees and shareholders of Bristol-Myers Squibb, Inc. For the remaining authors there are no conflicts of interest.
1. Guy GP, Machlin SR, Ekwueme DU, Yabroff KR. Prevalence and costs of skin cancer treatment in the U.S., 2002–2006 and 2007–2011 Am J Prev Med. 2015; 48: 183–187
2. Wolchok JD, Chiarion-Sileni V, Gonzalez R, Rutkowski P, Grob JJ, Cowey CL, et al. Overall survival with combined nivolumab and ipilimumab in advanced melanoma. N Engl J Med. 2017; 377: 1345–1356
3. Chapman PB, Hauschild A, Robert C, Haanen JB, Ascierto P, Larkin J, et al. Improved survival with vemurafenib in melanoma with BRAF V600E mutation. N Engl J Med. 2011; 364: 2507–2516
4. Hauschild A, Grob JJ, Demidov LV, Jouary T, Gutzmer R, Millward M, et al. Dabrafenib in BRAF-mutated metastatic melanoma: a multicentre, open-label, phase 3 randomised controlled trial. Lancet. 2012; 380: 358–365
5. Flaherty KT, Infante JR, Daud A, Gonzalez R, Kefford RF, Sosman J, et al. Combined BRAF and MEK inhibition in melanoma with BRAF V600 mutations. N Engl J Med. 2012; 367: 1694–1703
6. Hodi FS, O’Day SJ, McDermott DF, Weber RW, Sosman JA, Haanen JB, et al. Improved survival with ipilimumab in patients with metastatic melanoma. N Engl J Med. 2010; 363: 711–723
7. Larkin J, Chiarion-Sileni V, Gonzalez R, Grob JJ, Cowey CL, Lao CD, et al. Combined nivolumab and ipilimumab or monotherapy in untreated melanoma. N Engl J Med. 2015; 373: 23–34
8. Levit L, Balogh E, Lighter P, Nass S, Park M. Committee on Improving the Quality of Cancer Care: Addressing the Challenges of an Aging Population Institute of Medicine of the National Academies. 2013, Washington, DC: National Academies Press (US)
9. Lakdawalla DN, Romley JA, Sanchez Y, Maclean JR, Penrod JR, Philipson T. How cancer patients value hope and the implications for cost-effectiveness assessments of high-cost cancer therapies. Health Aff (Millwood). 2012; 31: 676–682
10. Lancsar E, Louviere J. Conducting discrete choice experiments to inform healthcare decision making. Pharmacoeconomics. 2008; 26: 661–677
11. Markman M, Luce R. Impact of the cost of cancer treatment: an internet-based survey. J Oncol Pract. 2010; 6: 69–73
12. Reed Johnson F, Lancsar E, Marshall D, Kilambi V, Mühlbacher A, Regier DA, et al. Constructing experimental designs for discrete-choice experiments: report of the ISPOR Conjoint Analysis Experimental Design Good Research Practices Task Force. Value Health. 2013; 16: 3–13
13. Krinsky I, Robb AL. On approximating the statistical properties of elasticities. Rev Econ Stat. 1986; 68: 715
14. Gogineni K, Shuman KL, Chinn D, Gabler NB, Emanuel EJ. Patient demands and requests for cancer tests and treatments. JAMA Oncol. 2015; 1: 33–39
15. Krammer R, Heinzerling L. Therapy preferences in melanoma treatment – willingness to pay and preference of quality versus length of life of patients, physicians and healthy controls. PLoS One. 2014; 9: e111237
16. Mühlbacher AC, Bethge S. Patients’ preferences: a discrete-choice experiment for treatment of non-small-cell lung cancer. Eur J Health Econ. 2015; 16: 657–670
17. Havrilesky LJ, Alvarez Secord A, Ehrisman JA, Berchuck A, Valea FA, Lee PS, et al. Patient preferences in advanced or recurrent ovarian cancer. Cancer. 2014; 120: 3651–3659
18. Postow MA, Chesney J, Pavlick AC, Robert C, Grossmann K, McDermott D, et al. Initial report of overall survival rates from a randomized phase II trial evaluating the combination of nivolumab (NIVO) and ipilimumab (IPI) in patients with advanced melanoma (MEL) Cancer Res. 2016; 76: CT002
19. Maio M, Grob JJ, Aamdal S, Bondarenko I, Robert C, Thomas L, et al. Five-year survival rates for treatment-naive patients with advanced melanoma who received ipilimumab plus dacarbazine in a phase III trial. J Clin Oncol. 2015; 33: 1191–1196
20. Long GV, Flaherty KT, Stroyakovskiy D, Gogas H, Levchenko E, Larkin J, et al. Dabrafenib plus trametinib versus dabrafenib monotherapy in patients with metastatic BRAF V600E/K-mutant melanoma: long-term survival and safety analysis of a phase 3 study. Ann Oncol. 2017; 28: 1631–1639
21. Slevin ML, Stubbs L, Plant HJ, Wilson P, Gregory WM, Armes PJ, et al. Attitudes to chemotherapy: comparing views of patients with cancer with those of doctors, nurses, and general public. BMJ. 1990; 300: 1458–1460
22. Welsh SJ, Rizos H, Scolyer RA, Long GV. Resistance to combination BRAF and MEK inhibition in metastatic melanoma: where to next? Eur J Cancer. 2016; 62: 76–85
23. Arozarena I, Wellbrock C. Overcoming resistance to BRAF inhibitors. Ann Transl Med. 2017; 5: 387