HM Medical Clinic

 

Long-term cost–effectiveness of oncotype dx^® versus current clinical practice from a dutch cost perspective

For reprint orders, please contact: [email protected] Long-term cost–effectiveness of Oncotype DX® versus current clinical practice from a Dutch cost perspective J. Comp. Eff. Res.
Introduction: This study analyzes the incremental cost–effectiveness of
Michelle Kip1,2, Helma Oncotype DX® testing to support adjuvant chemotherapy recommendations, versus Monteban3 & Lotte Steuten*,1,4 current clinical practice, for patients with estrogen receptor-positive (ER+), node- 1Panaxea BV, Health Economics negative or micrometastatic (pN1mic) early-stage breast cancer in The Netherlands. & Reimbursement, Enschede, Methods: Markov model projecting distant recurrence, survival, quality-adjusted
life years (QALYs) and healthcare costs over a 30-year time horizon. Results:
2Department of Health Technology Oncotype DX was projected to increase QALYs by 0.11 (0.07–0.58) and costs with & Services Research, University of Twente, Enschede, The Netherlands €1236 (range: -€142–€1236) resulting in an incremental cost–effectiveness ratio of 3Monteban Value Services, Amerongen, €11,236/QALY under the most conservative scenario. Conclusion: Reallocation of
adjuvant chemotherapy based on Oncotype DX testing is most likely a cost-effective 4Hutchinson Institute for Cancer use of scarce resources, improving long-term survival and QALYs at marginal or Outcomes Research (HICOR), Fred lower costs.
Hutchinson Cancer Research Center, Seattle, WA, USA *Author for correspondence: Approximately 14,000 women in The hand, this complexity is due to difficulties in Netherlands are diagnosed with breast can- capturing disease heterogeneity, incorporat- cer each year. Breast cancer remains the most ing tumor size, patient age and histology as common type of cancer in women and is risk factors [13–15]. On the other hand, adverse one of the most common causes of death for events related to chemotherapy treatment are women between the ages of 30 and 59 years experienced by almost all patients, varying in in The Netherlands [1,2]. Patients presenting severity from hair loss to death [13]. Besides with early-stage, estrogen receptor-positive the clinical pros and cons, the use of chemo- (ER+) breast cancer preferentially benefit from therapy is associated with high costs [16,17]. adjuvant endocrine therapy, as demonstrated The high rate of chemotherapy prescription in large randomized controlled trials [3,4]. The in early-stage ER+ breast cancer patients indi-benefit from chemotherapy in terms of pre- cates that a significant number of patients venting or delaying distant recurrence after might undergo chemotherapy treatment primary surgery, however, is uncertain and despite a low recurrence risk, and therefore present in a small group of patients with this have only a limited possibility of achieving type of cancer. Despite this, current Dutch the health benefits. Consequently, there is a consensus guidelines, based on standard need for a predictive tool to help guide the clinical-pathological criteria, still recom- decision to prescribe adjuvant chemotherapy, mend adjuvant chemotherapy plus endocrine which may reduce both the burden and cost therapy for the majority of these patients [5]. of breast cancer management [15–17].
Consequently, a complex decision concerning Recently, molecular assays to predict risk this patient category is whether to prescribe and treatment response for early stage breast adjuvant chemotherapy treatment in addi- cancer patients have become commercially tion to endocrine therapy [6–12]. On the one available [18]. Of these, the Oncotype DX® 10.2217/CER.15.18 2015 Future Medicine Ltd J. Comp. Eff. Res. (Epub ahead of print)
Research Article Kip, Monteban & Steuten assay has shown to provide both predictive and prog- more likely to follow the treatment suggested by Onco- nostic information, beyond traditional parameters, in type DX than low RS patients (RR: 1.07 [95% CI: node-negative (N0) and node-positive (N+) hormone 1.01−1.14]). While the clinical and patient benefits of receptor-positive disease [19–23]. This assay predicts both Oncotype DX are compelling, these have to be balanced the 10-year recurrence risk of breast cancer as well as against the costs of the Oncotype DX assay (€3180), and the expected benefit of adjuvant chemotherapy treat- the potential downstream cost impacts. This study ana- ment [2]. The recurrence score (RS) is expressed in a lyzed the incremental cost–effectiveness of Oncotype
numeric value between zero and 100 using a proprietary DX testing to support adjuvant chemotherapy decision-
algorithm. This score is used to stratify patients in three making versus current clinical practice in treatment of
risk groups: low, intermediate and high risk of disease patients with ER+, node-negative or micrometastatic
relapse [24]. The correlation between the Oncotype DX (pN1 mic) early-stage breast cancer in The Netherlands.
Recurrence Score (RS) and both distant recurrence and/
or survival, and the benefit from chemotherapy, has been Methods
reported by Paik et al. in two prospective analyses of Analytical framework
archived tissue from randomized clinical trials (NSABP A Markov model was developed using Microsoft Excel
B-14 and B-20) [19,20]. Results of this analysis indicated (Redmond, WA, USA) to model the recursive disease
that Oncotype DX independently predicts the 10-year processes, including the annual risk of recurrence and
recurrence risk of breast cancer, as well as the response mortality. The model is based on the original model
to adjuvant chemotherapy treatment [17,20]. Recently, structure by Hornberger et al. [18], which was initially
Carlson et al. (2013) reported that, based on a meta-
developed for a NICE submission in England and analysis of 21 studies, Oncotype DX changed overall Wales and has been adapted to the Dutch setting, in treatment recommendations in 33.4% of patients [25]. accordance with the applicable pharmacoeconomic They also found that high RS patients were significantly guidelines [26]. The analysis was performed from a Oncotype DX® testing Figure 1. Overview of model structure.
ESBC: Early-stage breast cancer.
J. Comp. Eff. Res. (Epub ahead of print)
future science group Cost–effectiveness of Oncotype DX® from a Dutch cost perspective Research Article Table 1. Summary of clinical input parameters.
SE (minimum;
Study (year)
Registration (2013) Net change in chemotherapy use -18.60 1.86 (-20.50; -9.30) 2.10 (-20.50; -10.48) 2.00 (-20.50; -10.00) 1.87 (-20.50; -9.34) Net change in chemotherapy use 7.5 0.75 (3.75; 11.25) with intermediate RS (%) 0.19 (0.95; 2.86) 0.05 (-0.75; -0.25) 0.32 (-4.82; -1.61) Net change in chemotherapy use 1.90 0.19 (0.95; 2.85) 0.48 (2.38; 7.14) 0.16 (0.80; 2.40) 0.44 (2.21; 6.63) 10-year risk of recurrence (low 1.60 (1.60; 4.80) RS) on endocrine therapy (%)10-year risk of recurrence 4.30 (4.55; 13.65) (intermediate RS) on endocrine therapy10-year risk of recurrence (high 7.30 (19.75; 59.25) RS) on endocrine therapy (%)RRR with chemotherapy N/A (0.00; 20.00) Assumption based RRR for chemotherapy 4.43 (19.5; 58.5) (intermediate RS) (%)RRR for chemotherapy 3.95 (37.0; 100.0) (high RS) (%)Postrecurrence survival (years) 0.330 (1.65; 4.95) Central Bureau of Statistics of The Netherlands (2012) HT: Hormone/endocrine therapy; N/A: Not applicable; RRR: Relative risk reduction; RS: Recurrence Score; SE: Standard error.
Dutch healthcare payer's perspective comparing the in the model start in the ‘recurrence-free' state, and clinical outcomes and costs of ‘Oncotype DX testing' in each 1-year cycle of the simulation, patients might to ‘current clinical practice' involving clinical–patho- experience either a recurrence or mortality event, and logical assessment, over a time horizon of 30 years. might therefore transition to either the ‘recurrence' or The model made projections of life expectancy, qual- the ‘dead' state (absorbing state). Transition probabili- ity-adjusted life expectancy and direct costs, based on ties between the three health states are based on Horn-recurrence rates for low-, intermediate- and high-risk berger et al. [27]. To avoid systematic over- or underes-patients as well as country-specific mortality data. The timation of survival in the model, half-cycle correction model structure is outlined in Figure 1.
was performed.
Three Markov health states are defined to estimate the cumulative costs and QALYs based on hypothetical Patient population & comparatorscohorts of 100 women with ER+, node-negative, or sin- Starting age of the cohort was 60 years, which is the gle node positive early-stage breast cancer. All patients median age of incident breast cancer cases as reported future science group Research Article Kip, Monteban & Steuten Table 2. Chemotherapy costs and adverse events.
50% patients 3× FEC 3× docetaxel treatment Mean 180 mg per patient per session (derived from 50% patients 6× TAC treatment regimen Taxotere/doxetaxel = 1.8 × 75 = 135 mg per session per patientAdriamycine/doxorubicine = 1.8 × 50 = 90 mg per session per patient G-CSF use with docetaxel G-CSF use with TAC Add-on DRG for docetaxel €4.07 per mg [36] Add-on DRG for doxorubicine €23.56 per mg [36] DRG for intravenous/intrathecal chemotherapy in 95% of patients (derived from expert), average tariff nonmetastatic tumors, nonclinical of €1225.64 as derived from publicly available data of 73 Dutch hospitals. Six DRGs per chemotherapy DRG for intravenous/intrathecal chemotherapy in 5% of patients (derived from expert), average tariff nonmetastatic tumors, with side effects requiring of €4409.91 as derived from publicly available data of hospitalization 73 Dutch hospitals. Six DRGs per chemotherapy DRG: Diagnosis-Related group; FEC: Fluorouracil, epirubicin, cyclophosphamide; TAC: Docetaxel, adriamycine, cyclophosphamide.
by The Netherlands Cancer Registration [28]. In cur- zero. All recurrence rates are assumed to be indepen- rent clinical practice, patients receive endocrine therapy dent of age. Nonbreast cancer death was captured as a and 3rd-generation chemotherapy regimens according competing risk in the model. These data were derived to Adjuvant! Online and the Nottingham Prognostic from The Netherlands Cancer Registration and based Index. Therapy recommendations in current clinical on Dutch female life tables for 2007–2009 [28,33]. For practice for low, intermediate and high risk were based patients experiencing distant recurrence, survival was on the study of Albanell et al. [29]. The net change assumed to be 3.3 years [32]. Long-term adverse events in the percentage of chemotherapy (CT) use follow- associated with chemotherapy are not captured in the ing Oncotype DX testing was also extracted from model (including cardiotoxicity, leukemia, and mild Albanell et al. for the base case analysis. In addition, cognitive impairment).
we modeled the net change in % CT use as reported in a German cohort (Eiermann et al. [30]), in a British Costscohort (Holt et al. [31]) and based on additional data The Markov model incorporates cost of endocrine from the Carlson et al. [25] meta-analysis as provided therapy, cost of chemotherapy and cost of distant by the authors. Data input for the base case (Albanell) recurrence. Cost of endocrine and chemotherapy, and and alternative analyses (Eiermann, Holt and Carlson) cost of accompanying adverse events, were derived are specified in Table 1 [25,29–31].
from tariffs of The Netherlands Healthcare Authority (NZA) and expressed in Euros [34]. Costs of endocrine therapy were derived from the NZA and from the Data on recurrence risks associated with endocrine official website of the National Health Care Institute, therapy and relative risk reduction associated with listing medication costs [35]. Endocrine treatment chemotherapy, were derived from the NSABP B-20 costs are incurred over 8 years, with higher annual cohort. These data were evaluated for each simulated costs in years 1–5 and lower costs in years 6–8, reflect-patient in each cycle of the model, based on their RS ing change in treatment intensity over time. One (low, intermediate, or high), as shown in Table 1 [20]. hundred percent therapy compliance was assumed. These risks were adjusted based on whether patients The treatment cost of the two most commonly used were receiving chemotherapy as per the initial recom- third-generation CT regimens were applied assum- mendation (in the current clinical practice arm) and ing a 50–50 distribution among all patients recom-based on the RS (in the Oncotype DX arm). Relative mended for CT, as based on communication from risk reduction of distant recurrence among low risk physicians of the Leiden University Medical Center in patients receiving chemotherapy was assumed to be The Netherlands. The first regimen involves six cycles J. Comp. Eff. Res. (Epub ahead of print)
future science group Cost–effectiveness of Oncotype DX® from a Dutch cost perspective Research Article of TAC (T = docetaxel, A = adriamycine, C = cyclo- rence. Thomas et al. stated that the population cur- phosphamide), combined with primary G-CSF ther- rently under consideration (ER+ and HER2-), incurred apy. The second one involves three cycles of FE100C lower treatment costs than ER+ and HER+ patients. (5-fluorouracil 500 mg/m2, epirubicin 100 mg/m2, While their estimation of mean cost across all groups cyclophosphamide 500 mg/m2), and three cycles of may thus be an overestimate for this population, it is docetaxel. For those patients receiving chemotherapy, still lower than that reported elsewhere [38]. Follow-the costs of chemotherapy and endocrine therapy are ing Dutch pharmacoeconomic guidelines, future costs both incurred in the first year, indicating an overlap were discounted at 4% and clinical benefits at 1.5% of costs of approximately 3.5 months. Costs of che- annually. A summary of the cost variables used is motherapy and G-CSF are included in the relevant provided in Table 3.
diagnosis related group (DRG) tariff, which further includes all in- and out-patient activities that a hospi- tal performs following an ICD-10 diagnosis as well as Health utility values were obtained from published lit-the honorarium for the medical specialists involved. erature and were used to calculate the quality-adjusted Some medications are not included in the DRGs and, life-years (QALYs). These scores are quantitative rep-therefore, considered as DRG add-ons. In this case, resentations of the desirability of a particular health costs of docetaxel and doxorubicin are add-ons to outcome, with a utility of 0 being equivalent to death the DRG, and costs have been derived for an average and a value of 1 equaling perfect health. Health-related patient (135 mg docetaxel, 90 mg doxorubicin) [36]. quality of life associated with recurrence is taken from A summary of all costs associated with chemotherapy Milne, who reported an analysis in New Zealand costs, its adverse events and the corresponding data women with advanced breast cancer, and assumed treat-sources is provided in Table 2.
ment via endocrine therapy [39]. In addition, a disutility Other drug costs were derived from the Pharma- associated with chemotherapy of 0.07 was applied to cotherapeutic Compass [37]. Adverse events associ- capture the health-related quality of life impact of che- ated with endocrine therapy are effectively set to zero, motherapy in the first model cycle (for those patients because these costs are incorporated in the DRG tariff. recommended chemotherapy in each treatment arm), Costs of distant recurrence were derived from a study and this value was taken from Peasgood et al. [40]. The by Thomas et al. including active, supportive and end health utility associated with one year in the recurrence of life care, and recalculated to 2012 Euros [32]. Total free state was assumed to be the same during and after cost of recurrence is incurred at the time of recur- endocrine therapy (Connor-Spady et al.) [41]. The util- Table 3. Summary of cost variables in the cost–effectiveness modeling analysis.
Mean cost (€) SE
Oncotype DX® test Genomic Health (2014 list Endocrine therapy (years 1–5) 877.71 DRG tariff derived from the NZa and medicijnkosten.
Endocrine therapy (years 6–8) 114.27 DRG tariff derived from the NZa and medicijnkosten.
DRG tariff derived from the NZa and Pharmacotherapeutic Distant recurrence (monthly) Thomas et al. (2009) [32] Endocrine therapy adverse Not taken into account, events (years 1–5) Endocrine therapy adverse Not taken into account, events (years 6–8) Chemotherapy adverse events 0.01 Not taken into account, SE: Standard error.
future science group Research Article Kip, Monteban & Steuten Table 4. Utility scores used in the modeling analysis.
State or event
Mean utility
Distribution Study (year)
1 year in recurrence-free state Conner-Spady (2005) 1 year in recurrence state Chemotherapy treatment (six cycles) -0.07 SE: Standard error.
ity scores used in the modeling analysis are reported minimum values are assumed at ± 50% of the mean. in Table 4.
These assumptions were considered conservative since they are likely to exceed the true variance. The appro- Sensitivity analyses priate number of runs to be included in the PSA was One-way and probabilistic sensitivity analyses (PSA) estimated based on the mean ICER plotted against the were conducted. In the one-way sensitivity analy- number of iterations of the PSA. Variation in sampling sis, key drivers of model outcome were identified was found to be sufficiently minimized when running by varying each of the individual input parameters the PSA 1000 times.
over its predetermined range. Both costs and clini- The model was validated by a third party health eco- cal parameters were varied, and the impact of each nomic modeler, who reviewed both the face-validity of on the incremental cost–effectiveness ratio (ICER) the model structure and the formulas used. No further was noted. The impact of five (a priori determined) sensitivity analyses were performed to investigate struc-key-parameters will be shown in a tornado diagram, tural uncertainty. An internal validation exercise was namely the mean starting age of the patient cohort, performed, to compare model outcomes against the the cost of Oncotype DX, the net change in chemo- study on which it was based. Results from the NICE therapy use following Oncotype DX testing, the ten model and the validation exercise were found to closely
year risk of recurrence with hormonal therapy, the match in both treatment arms (data not shown).
relative risk reduction (RRR) of recurrence with use
of chemotherapy. For the latter three parameters, the Results
sensitivity is evaluated for low, intermediate and high Oncotype DX testing was associated with a notable
risk separately.
change in treatment recommendations, of which our The joint decision uncertainty surrounding the base case reflects the most conservative scenario with model outcomes was quantified by means of a proba- 18.6% less CT use in low risk patients, and 7.5% bilistic sensitivity analysis (PSA), which employs ran- and 1.8% more CT use in intermediate and high dom draws from predetermined parameter distribu- risk patients, respectively (Table 5). The data from tions that were defined using the method-of-moments Eiermann et al. reflect the scenario with the highest approach [42]. Where reported, distribution parameters impact of Oncotype DX on the net change in CT use, (such as standard error [SE] and minimum and maxi- with -20, -0.5 and +1.6% for low-, intermediate- and mum values) were taken from literature. When insuf- high-risk patients, respectively [30].
ficient information was available, the standard error A scatter plot presenting the base case results from was assumed to be 10% of the mean; maximum and the PSA is shown in Figure 2, depicting the incremen- Table 5. Summary of changes in adjuvant therapy recommendations with Oncotype DX®.
Recurrence score
Post Oncotype® DX
Net change CT (%)
ET + CT (%)
ET + CT (%)
CT: Chemotherapy; ET: Endocrine therapy.
Data taken from [29].
J. Comp. Eff. Res. (Epub ahead of print)
future science group Cost–effectiveness of Oncotype DX® from a Dutch cost perspective Research Article tal costs and QALYs for a cohort of 100 patients. The tive compared with current clinical practice in The mean difference in quality-adjusted life expectancy for Netherlands. Oncotype DX testing was associated Oncotype DX versus current clinical practice is 0.11 with a notable change in treatment recommenda-QALYs per patient, and comes at an incremental cost tions based on the data reported in three different of €1236 per patient, yielding an ICER of €11,236 per original studies and a meta-analysis. Based on these QALY gained. A cost–effectiveness acceptability curve changes, and modeling the long-term risk of distant was plotted based on the findings of the PSA with will- recurrence based on published trial data, Oncotype ingness to pay thresholds ranging from €0 per QALY DX was projected to improve quality-adjusted life gained to €40,000 per QALY gained. At a willingness expectancy per patient with 0.11 QALYs and 0.13 life to pay threshold of €30,000 per QALY there is a 99.5% years in the base case analysis, which is also the most probability that Oncotype DX is cost effective versus conservative.
current clinical practice (data not shown).
PSA showed that there was a 99.5% probability that Table 6 shows per patient costs, QALYs, life years Oncotype DX would be cost effective, at a willingness (LYs) and ICERs for the base case and the alterna- to pay threshold of €30,000 per QALY gained. One-way tive analyses. Incremental costs are negative in the sensitivity analyses showed that the cost–effectiveness analysis based on Eiermann et al. data [30] (-€124 per of Oncotype DX testing was most sensitive to varia-patient) and positive in the three other analyses (range: tions in patient age, cost of the Oncotype DX assay and €325–1236). In all analyses, incremental QALYs and net changes in chemotherapy use in low risk patients. LYs are higher for the Oncotype DX versus the current Regarding the latter, it is notable that the studies under-clinical practice strategy, varying from +0.07 QALYs pinning this model are very consistent regarding the and +0.07 LYs (Eiermann et al. [30]) to 0.58 QALYs percentage change in treatment recommendation in and +0.73 LYs (Holt et al. [31]). The ICERs range the total patient group (varying between 27% [31] and from +€11.236/QALY and +€9.508/LY gained 33% [25,30]), but less so regarding the net change in che-(based on Albanell et al. [29]) to dominance (based on motherapy in the intermediate- and high-risk groups. Eiermann et al. [30]).
While we cannot provide a full explanation for this The tornado diagram in Figure 3 shows that varying based on the reported data, we observe that a positive the starting age of the patient cohort has the highest net change in chemotherapy in the intermediate-risk impact on the model outcome, followed by the cost group [25,30], and a higher positive net change in the of the Oncotype DX assay, and the net change in high-risk group [25,31], coincides with a relatively higher chemotherapy use following Oncotype DX testing.
proportion of patients classified in each of these groups. That said, in all sensitivity analyses, Oncotype DX was Discussion & conclusion
associated with ICERs in the range (€20,000–€80,000) This cost–effectiveness analysis contributes to the that would be considered cost effective by commonly evidence that Oncotype DX is likely to be cost effec- quoted standards in The Netherlands.
Cost (€
Figure 2. Incremental cost–effectiveness scatterplot.
CE: Cost–effectiveness; ICER: Incremental cost–effectiveness ratio; QALY: Quality-adjusted life year.
future science group Research Article Kip, Monteban & Steuten Table 6. Summary of cost–effectiveness results for the four studies.
Current clinical Oncotype
DX® testing
Quality-adjusted life expectancy (QALYs) Albanell Life expectancy (years) ICER (cost per QALY gained) ICER (cost per life year gained) Values shown are per patient.
ICER: Incremental cost–effectiveness ratio; QALY: Quality-adjusted life year.
Data taken from [25,29–31].
The cost–effectiveness analysis can be considered tions have been conservative (overestimating the likely conservative in some regards. First, this analysis did variance). Although sensitivity analyses have not been not capture some of the long-term adverse effects of performed to investigate the impact of changing these chemotherapy treatment (including cardiotoxicity, sec- assumptions, it is unlikely that any changes within ondary leukemia or effects on cognitive impairment). plausible limits would notably alter the findings of Local recurrence was not captured in the modeling the analysis, although specific results and estimates of analysis and, given the purported benefits in terms of uncertainty may be influenced.
treatment recommendations with Oncotype DX test- Although recommending a similar study in the ing, this may also have led to an underestimation of the Dutch setting, comparing Oncotype DX versus current clinical benefit of Oncotype DX.
clinical practice, might seem obvious, the added value A potential weakness of the analysis is that, in of such research is questionable. First of all, the results absence of Dutch data sources, it had to rely on long- of this analysis were generally consistent with those term clinical data from the USA to estimate the risk of observed in multiple other countries [43,44], with dif-distant recurrence. Also, health-related quality of life ferent clinical practices and other cost levels. Second, utility scores were not Dutch-specific. Ideally, clini- the high costs associated with performing those stud- cal data from a Dutch setting addressing both of these ies, and the inevitable burden to patients for participat-aspects would have been available to support the mod- ing in a study, are serious considerations to be made. eling analysis. Furthermore, information on variance Although no formal Value of Information analysis has around many of the model inputs was limited and, been performed, the current analysis shows little deci-therefore, assumptions have been made regarding the sion uncertainty and the one-way sensitivity analyses shape and, in certain cases, the parameters defining indicate that other values for key-inputs are unlikely distributions for PSA. In most cases, these assump- to change the overall recommendation. Instead, with J. Comp. Eff. Res. (Epub ahead of print)
future science group Cost–effectiveness of Oncotype DX® from a Dutch cost perspective Research Article more prognostic tests coming to the market nationally come the challenges mentioned above. Yet, in order to and internationally, for example, the Mammaprint®, base pricing on value, it is necessary that the definition MammaTyper®, PAM50, Endopredict and others [45], of value also captures the benefits of diagnostics appro-head-to-head studies regarding their impact on treat- priately [48]. The implementation of pricing flexibility ment decisions would provide important and timely could be facilitated by risk-sharing schemes, such as information for appropriate, value-based, adoption Oncotype DX and the US payer UnitedHealthcare and reimbursement decisions. Furthermore, while the agreed to, whereby they collected information on Oncotype DX is furthest along the validation pathway whether women actually forego chemotherapy when compared with the other genetic tests mentioned above, recommended to do so by the diagnostic.
the need remains for evaluating the accuracy of the Finally, diagnostic reimbursement is in many coun- results and predictions of the proprietary algorithms of tries, including The Netherlands, not always included these tests before definitive conclusions can be drawn in the cost of care for outpatients, and is commonly on any of them. Prospective studies are needed to do spread across multiple siloed budgets. This means that so, while longitudinal (observational) studies can help a test developer may face high barriers to getting a test to understand the relationship between predicted risk adopted because of budget constraints and costs falling and observed risk and hence its real-word impact on on other budget silos than those where the gains are clinical outcomes [46].
enjoyed, even when it has good evidence of cost–effec- When recommending more research, urgent atten- tiveness [49]. Detailed budget impact analyses, which tion is also required to the situation that payments for project the potential financial impact of the adoption diagnostics do not reflect the cost of evidence genera- and diffusion of a new technology into a healthcare tion or the value created [47]. Currently, payments to system with finite resources, are a means to generate drug manufacturers are typically based on projections insight in which budget holder is paying for the diag-of likely impacts on QALYs and costs, while diagnos- nostic and which one absorbs the downstream cost sav- tic reimbursement tends to be based on cost only. As ings [50]. Potential disincentives can herewith be identi-a consequence, neither the full benefits nor costs of fied and when budget holders are (held) accountable effective stratification are fully reflected in the pricing they could use this information to align the incentives and reimbursement systems for drugs and diagnostics and find a way to adopt a highly cost-effective diagnos-in most EU healthcare systems including The Nether- tic. In The Netherlands, health insurers are particularly lands. A system of flexible pricing has been proposed by well positioned to play a leading role in this, as they the Academy of Medical Sciences (2013) that is based negotiate prices for each Diagnostic Related Group on value both for drugs and for diagnostics, to over- with hospitals and primary care groups on an annual ICER (cost per QALY)
Net change in CT use following Oncotype DX® testing (low) Net change in CT use following Oncotype DX testing (intermediate) Net change in CT use following Oncotype DX testing (high) 10-year risk of recurrence with HT (low) 10-year risk of recurrence with HT (intermediate) 10-year risk of recurrence with HT (high) RRR of recurrence with use of CT (low) RRR of recurrence with use of CT (intermediate) RRR of recurrence with use of CT (high) Mean starting age of patient cohort Figure 3. Tornado diagram of one-way sensitivity analyses.
CT: Chemotherapy; HT: Hormone/endocrine therapy; ICER: Incremental cost–effectiveness ratio; QALY: Quality-adjusted life year;
RRR: Relative risk reduction.
future science group Research Article Kip, Monteban & Steuten basis. Given they are one of the major beneficiaries of a flexible approach to evidence generation, which bal-the significant cost savings due to expensive and inef- ances the need for high-quality evidence with appro- fective chemotherapies avoided by diagnostics such as priate incentives for innovation, will receive more Oncotype DX, they have generally been unresponsive attention in the future. Adaptive or observational tri-as to play a constructive role in this debate.
als designs and the application of advanced analytics to real-world data may well be the way forward, and there is large potential for big data analytics to sup- The Oncotype DX test is currently considered for port such comparative effectiveness studies. As such, reimbursement in The Netherlands and other EU Health Technology Assessment bodies, regulators countries. In the USA, it is already included in major and payers must (re-)consider not only the evidence treatment guidelines for breast cancer and it receives a requirements for informed decision-making, but also value-based reimbursement, which is based on clinical the consequences of such demands to healthcare data demonstrating the test's ability to restrict health- innovation, sustainability and population health.
care costs. The test is often considered the poster child for other genomic tests that are not linked to one spe- cific drug and herewith encounter the typical drug- The authors grateful y acknowledge Dr Roth of the Fred diagnostic co-development model. Notwithstanding Hutchinson Cancer Research Institute and Dr Carlson of the the appeal of such tests, most countries, except the University of Washington (Seattle, WA, USA) to share the UK, do not currently have a framework for calculat- crude data of their previously published systematic review ing the added value of stratification and separating the and meta-analysis [25] for reanalysis in this cost–effectiveness rewards for this between the therapeutic and diagnos- tic part [47]. Another current challenge is generating the evidence base for the clinical utility and predictive Disclaimervalue of these tests. While Oncotype DX and some All views expressed in this paper are the author's and the other genetic assays have been shown to effectively author's only.
guide treatment decisions, the use of such prognos-tic tests is not necessarily predictive, and long-term Financial & competing interests disclosurepatient outcomes are still unknown. Generating such Monteban Value Services b.v. and PANAXEA b.v. received evidence, requires large prospective studies that take funding from Genomic Health Inc. for developing and ana-time and substantial investments (from often fairly lyzing the model for The Netherlands. The authors have no small diagnostics companies) to be performed. Given other relevant affiliations or financial involvement with any the speed at which technology as well as usual care organization or entity with a financial interest in or financial changes for this sector, a strict insistence on random- conflict with the subject matter or materials discussed in the ized control trials, for instance, is increasingly unfea- manuscript apart from those disclosed.
sible and likely leaves a new diagnostic outdated by No writing assistance was utilized in the production of this the time premarket evidence is generated. Therefore, manuscript.
Executive summary Rationale
• Oncotype DX® is likely to be cost effective compared with current clinical practice in The Netherlands.
• The assay is expected to improve survival and quality-adjusted life expectancy.
Findings
• The assay comes at an incremental costs of €1236 per patient, yielding an ICER of €11,236 per QALY gained.
• The cost–effectiveness of testing is most sensitive to variations in patient age, assay cost and net changes in
chemotherapy use in low-risk patients.
• The results of this analysis were consistent with those observed in multiple other countries.
Discussion
• Even cost-effective diagnostics face high barriers to adoption and reimbursement because their costs and
benefits fall on separate budget silos.
• Currently, neither the full benefits nor costs of effective patient stratification are fully reflected in the pricing and reimbursement systems for drugs and diagnostics.
Recommendation
• Future studies should focus on head-to-head comparisons of various assays to provide timely information for
adoption and reimbursement decisions.
J. Comp. Eff. Res. (Epub ahead of print)
future science group Cost–effectiveness of Oncotype DX® from a Dutch cost perspective Research Article nonresponsive tumors: results from the National Surgical Papers of special note have been highlighted as: Adjuvant Breast and Bowel Project B-15. J. Clin. Oncol. 8(9), • of interest; •• of considerable interest 1483–1496 (1990).
Voogd AC. Borstkanker samengevat. Presented at: Fisher B, Dignam J, Wolmark N et al. Tamoxifen and Volksgezondheid Toekomst Verkenning. Nationaal Kompas chemotherapy for lymph node-negative, estrogen receptor- Volksgezondheid; RIVM (2013). positive breast cancer. J. Natl Cancer Inst. 89(22), 1673–1682 Klang SH, Hammerman A, Liebermann N, Efrat N, Dowsett M, Goldhirsch A, Hayes DF, Senn HJ, Wood W, Doberne J, Hornberger J. Economic implications of 21-gene Viale G. International Web-based consultation on priorities breast cancer risk assay from the perspective of an Israeli- for translational breast cancer research. Breast Cancer managed health-care organization. Value Health 13(4), Res. 9(6), R81 (2007).
381–387 (2010).
Barron JJ, Quimbo R, Nikam PT, Amonkar MM. Assessing Burstein HJ, Prestrud AA, Seidenfeld J et al. American the economic burden of breast cancer in a US managed Society of Clinical Oncology clinical practice guideline: care population. Breast Cancer Res. Treat. 109(2), 367–377 update on adjuvant endocrine therapy for women with hormone receptor-positive breast cancer. J. Clin. Broekx S, Den Hond E, Torfs R et al. The costs of breast Oncol. 28(23), 3784–3796 (2010).
cancer prior to and following diagnosis. Eur. J. Health This paper presents evidence-based guidelines, based
Econ. 12(4), 311–317 (2011).
on a systematic review, for endocrine therapy for
Hornberger J, Cosler LE, Lyman GH. Economic analysis of postmenopausal women with hormone receptor-positive
targeting chemotherapy using a 21-gene RT-PCR assay in breast cancer. It finds that an adjuvant treatment strategy
lymph-node-negative, estrogen-receptor-positive, early-stage incorporating an aromatase inhibitor as primary,
breast cancer. Am. J. Manag. Care 11(5), 313–324 (2005).
sequential, or extended therapy reduces the risk of breast
Paik S, Shak S, Tang G et al. A multigene assay to predict cancer recurrence compared with 5 years of tamoxifen
recurrence of tamoxifen-treated, node-negative breast cancer. N. Engl. J. Med. 351(27), 2817–2826 (2004).
Early Breast Cancer Trialists' Collaborative Group. Paik S, Tang G, Shak S et al. Gene expression and benefit Polychemotherapy for early breast cancer: an overview of chemotherapy in women with node-negative, estrogen of the randomised trials. Early Breast Cancer Trialists' receptor-positive breast cancer. J. Clin. Oncol. 24(23), Collaborative Group. Lancet 352(9132), 930–942 (1998).
3726–3734 (2006).
Nabon: Landelijke richtlijn mammacarcinoom 2012. Habel LA, Shak S, Jacobs MK et al. A population-based study of tumor gene expression and risk of breast cancer death among lymph node-negative patients. Breast Cancer Muss HB. Adjuvant therapy for older women with breast cancer. Breast 12(6), 550–557 (2003).
Res. 8(3), R25 (2006).
Dowsett M, Cuzick J, Wale C et al. Prediction of risk of Bergh J. Best use of adjuvant systemic therapies II, chemotherapy aspects: dose of chemotherapy-cytotoxicity, distant recurrence using the 21-gene recurrence score in duration and responsiveness. Breast 12(6), 529–537 (2003).
node-negative and node-positive postmenopausal patients with breast cancer treated with anastrozole or tamoxifen: a Cardoso F, Piccart MJ. The best use of chemotherapy in the TransATAC study. J. Clin. Oncol. 28(11), 1829–1834 (2010).
adjuvant setting. Breast 12(6), 522–528 (2003).
Albain KS, Barlow WE, Shak S et al. Prognostic and Pritchard KI. The best use of adjuvant endocrine treatments. predictive value of the 21-gene recurrence score assay in Breast 12(6), 497–508 (2003).
postmenopausal women with node-positive, oestrogen- Piccart MJ, Sotiriou C, Cardoso F. New data on receptor-positive breast cancer on chemotherapy: a chemotherapy in the adjuvant setting. Breast 12(6), 373–378 retrospective analysis of a randomised trial. Lancet Oncol. 11(1), 55–65 (2010).
Goldhirsch A, Wood WC, Coates AS, Gelber RD, Lamond NW, Skedgel C, Younis T. Is the 21-gene recurrence Thurlimann B, Senn HJ. Strategies for subtypes-dealing with score a cost-effective assay in endocrine-sensitive node- the diversity of breast cancer: highlights of the St. Gallen negative breast cancer? Expert Rev. Pharmacoecon. Outcomes International Expert Consensus on the Primary Therapy of Res. 13(2), 243–250 (2013).
Early Breast Cancer 2011. Ann. Oncol. 22(8), 1736–1747 Carlson JJ, Roth JA. The impact of the Oncotype Dx breast cancer assay in clinical practice: a systematic review and meta- Carlson RW, Allred DC, Anderson BO et al. Breast cancer. analysis. Breast Cancer Res. Treat. 141(1), 13–22 (2013).
Clinical practice guidelines in oncology. J. Natl Compr. •• A systematic review and meta-analysis were conducted to
Cancer Netw. 7(2), 122–192 (2009).
help stakeholders navigate and interpret the Oncotype DX®
Fisher B, Brown AM, Dimitrov NV et al. Two months literature. An important finding is that the treatment
of doxorubicin-cyclophosphamide with and without decision suggested by Oncotype Dx as well as its clinical
interval reinduction therapy compared with 6 months and economic impact is most clear for the high and low-risk
of cyclophosphamide, methotrexate, and fluorouracil in positive-node breast cancer patients with tamoxifen- future science group Research Article Kip, Monteban & Steuten Hakkaart-Van Roijen L, Tan SS, Bouwmans CaM. Nederlandse Zorg Autoriteit: Bijlage 2 bij Beleidsregel CU- [Guideline for costing research: methods and standardized 2065 Tarieflijst Instellingen 2012. costprices for economic evaluations in health care]. Diemen, College voor Zorgverzekeringen (2010).
Zorginstituut Nederland: Medicijnkosten 2012. Hornberger J, Alvarado MD, Rebecca C, Gutierrez HR, Yu TM, Gradishar WJ. Clinical validity/utility, change Nederlandse Zorg Autoriteit: DBC zorgproducten in practice patterns, and economic implications of risk tariefapplicatie 2014. stratifiers to predict outcomes for early-stage breast cancer: a systematic review. J. Natl Cancer Inst. 104(14), 1068–1079 (2012).
College Voor Zorgverzekeringen: Farmacotherapeutisch Kompas. 2013. Integraal Kankercentrum Nederland: Cijfers over kanker 2013. www.iknl.nl/ Remak E, Brazil L. Cost of managing women presenting with stage IV breast cancer in the United Kingdom. Br. J. Albanell J, Gonzalez A, Ruiz-Borrego M et al. Prospective Cancer 91(1), 77–83 (2004).
transGEICAM study of the impact of the 21-gene Recurrence Score assay and traditional clinicopathological Milne RJ, Heaton-Brown KH, Hansen P, Thomas D, Harvey V, Cubitt A. Quality-of-life valuations of advanced breast factors on adjuvant clinical decision making in women with cancer by New Zealand women. estrogen receptor-positive (ER Pharmacoeconomics 24(3), +) node-negative breast cancer. 281–292 (2006).
Ann. Oncol. 23(3), 625–631 (2012). Peasgood T, Ward SE, Brazier J. Health-state utility values in •• This study examines the impact of the Recurrence
breast cancer. Expert Rev. Pharmacoecon. Outcomes Res. 10(5), Score in Spanish breast cancer patients and explored
553–566 (2010).
the associations between clinicopathological markers
and likelihood of change in treatment recommendation.
Conner-Spady BL, Cumming C, Nabholtz JM, Jacobs Results from this first prospective European study are
P, Stewart D. A longitudinal prospective study of health-related quality of life in breast cancer patients following consistent with published experience and provide further
high-dose chemotherapy with autologous blood stem cell evidence on how Oncotype DX and clinicopathological
transplantation. Bone Marrow Transplant. 36(3), 251–259 factors are complementary and patient selection may be
Briggs A, Claxton K. Decision Modelling for Health Eiermann W, Rezai M, Kummel S et al. The 21- Economic Evaluation. Oxford University Press, Oxford, UK gene recurrence score assay impacts adjuvant therapy recommendations for ER-positive, node-negative and node-positive early breast cancer resulting in a risk-adapted Rouzier R, Pronzato P, Chéreau E, Carlson J, Hunt B, change in chemotherapy use. Ann. Oncol. 24(3), 618–624 Valentine WJ. Multigene assays and molecular markers in breast cancer: systematic review of health economic analyses. Breast Cancer Res. Treat. 139(3), 621–637 (2013).
•• A prospective clinical study was done in Germany to
evaluate the impact of the Recurrence Score on treatment
This systematic review examines published health
decisions in early breast cancer. Overall, 33% fewer
economic analyses to assess the support for adjuvant
patients actually received chemotherapy as compared with
therapy decision-making. Evidence suggests that multigene
patients recommended chemotherapy pre-test. Using the
assays are likely cost saving or cost effective relative to
test was cost-saving versus current clinical practice.
current approaches to adjuvant therapy and should benefit
decision-making in early-stage breast cancer in a variety of
Holt S, Bertelli G, Humphreys I et al. A decision impact, decision conflict and economic assessment of routine Oncotype DX testing of 146 women with node-negative Kondo M, Hoshi SL, Yamanaka T, Ishiguro H, Toi M. or pNlmi, ER-positive breast cancer in the UK. Br. J. Economic evaluation of the 21-gene signature (Oncotype Cancer 108(11), 2250–2258 (2013).
DX) in lymph node-negative/positive, hormone receptor-positive early-stage breast cancer based on Japanese The Oncotype DX test is evaluated for the NHS including
validation study (JBCRG-TR03). Breast Cancer Res. its impact on costs and treatment recommendations by
Treat. 127(3), 739–749 (2011).
UK oncologists. Findings suggest that Oncotype Dx is
Sinn P, Aulmann S, Wirtz R et al. Multigene assays for applicable and feasible to perform in UK patients with a
classification, prognosis, and prediction in breast cancer: reduction in the use of adjuvant chemotherapy, consistent
a critical review on the background and clinical utility. with findings of other reported studies.
Geburtshilfe. Frauenheilkd. 73(9), 932–940 (2013).
Thomas RJ, Williams M, Marshall C, Glen J, Callam M. Marchionni L, Wilson RF, Wolff AC et al. Systematic The total hospital and community UK costs of managing review: gene expression profiling assays in early-stage breast patients with relapsed breast cancer. Br. J. Cancer 100(4), cancer. Ann. Intern. Med. 148(5), 358–369 (2008).
598–600 (2009).
The Academy of Medical Sciences. Realising the potential of Central Bureau of Statistics (Nl). Mortality rates by age and stratified medicine. 1–106 (2013). gender (2012).
J. Comp. Eff. Res. (Epub ahead of print)
future science group Cost–effectiveness of Oncotype DX® from a Dutch cost perspective Research Article Garau M, Towse A, Garrison L, Housman L, Ossa D. Can Miller I, Ashton-Chess J, Spolders H et al. Market access and should value based pricing be applied to molecular challenges in the EU for high medical value diagnostic tests. diagnostics? OHE Research paper 12/03 (2012).
Personal. Med. 8(2), 137–148 (2011).
•• The purpose of this paper is to discuss how a value-based
Sullivan SD, Mauskopf JA, Augustovski F et al. Budget pricing framework for efficient use and pricing of medicines
impact analysis-principles of good practice: report of the also might be applied to diagnostics. Assessment of
ISPOR 2012 Budget Impact Analysis Good Practice II Task diagnostics should consider a broad range of value elements
Force. Value Health 17(1), 5–14 (2014).
and be based on a balanced analysis of diagnostic impact.
future science group

Source: https://www.panaxea.eu/wp-content/uploads/2015/04/Kip-et-al.-JCompEffRes04152015epub1.pdf