A cost-consequence model of using the 21-gene assay to identify patients with early-stage node-positive breast cancer who benefit from adjuvant chemotherapy in the Netherlands

Figures & data

Figure 1. Structural overview of the cost-consequence model. Total costs were calculated for every testing strategy. Costs for ACT and short-term AE management are calculated over a period of 6 months and for treating recurrent disease and long-term AE management over a period of 5 years. Abbreviations. ACT, adjuvant chemotherapy; AE, adverse events; GEP, gene expression profiling; HER2–, human epidermal growth factor receptor 2 negative; HR+, hormone receptor positive; N1, nodal stage (1–3 positive lymph nodes).

Table 1. Model input parameters.

Parameter	Value	Source
Population, n (%)
Patients with early-stage invasive breast cancer	15,000 (100.0)	^5
Patients in disease stage I	3,015 (20.1)	^17
Patients in disease stage II	6,630 (44.2)	^17
Patients in disease stage IIIa^a	1,605 (10.7)	^17
Patients with HR+/HER2–	11,025 (73.5)	^17
Patients with N1	2,205 (14.7)	^17
Post-menopausal women eligible for ACT^b	7,284 (48.6)	^5
Total eligible population in the model^c	590 (3.9)	^5^,^17
Proportions of risk scores
Oncotype DX assay result = 0–13	0.371	^1
Oncotype DX assay result = 14–25	0.458	^1
Oncotype DX assay result >25^d	0.171	^1
MammaPrint – low risk^e	0.695	^9
MammaPrint – high risk^e	0.305	^9
No GEP testing – high risk^f	1.000	EO
Five-year recurrence rates per risk group
Oncotype DX assay result = 0–13	0.073	^1
Oncotype DX assay result = 14–25	0.087	^1
Oncotype DX assay result > 25	0.164	^18
MammaPrint – low risk	0.084	^9
MammaPrint – high risk	0.141	^9
No GEP testing – high risk^g	0.095	^1^,^18
Proportion of patients across the treatment regimens
Regimen 1: 3 × FEC100 + 3 × T	0.500	^6^,^19
Regimen 2: 4×(dd)AC + 12 × P (weekly)	0.250	^6^,^19
Regimen 3: 4×(dd)AC + 4 × T	0.250	^6^,^19
GEP assay costs (per patient), €
Oncotype DX	4,594	^21
MammaPrint	3,863	^22
Drug acquisition costs (per patient), €^h
Regimen 1: 3 × FEC100 + 3 × T	5,016	^25
Regimen 2: 4×(dd)AC + 12 × P (weekly)	6,492	^25
Regimen 3: 4×(dd)AC + 4 × T	5,276	^25
Drug administration costs (per patient), €^i
Regimen 1: 3 × FEC100 + 3 × T	5,075	^23
Regimen 2: 4×(dd)AC + 12 × P (weekly)	11,600	^23
Regimen 3: 4×(dd)AC + 4 × T	5,800	^23
Costs for treating short-term AE (weighted average of all short-term AEs per patient), €	3,876	^26–30
Costs for treating long-term AE (weighted average of both long-term AEs per patient), €	589	^31–34
Productivity losses (average per patient), €	5,524	^5^,^35–37
Cost for treating a recurrence (average per patient), €	54,002	^18^,^38

^a It was assumed that 50% of the n = 3,210 patients with stage III would have stage IIIa.

^b It was assumed that on average women were post-menopausal when they were 50 years old and still eligible for receiving ACT until they were 70 years old.

^c The total amount of eligible patients for the model was calculated using the proportions of the different subgroups of the total population: 15,000 * (0.201 + 0.442 + 0.107) * 0.735 * 0.147 * 0.486 = 590.

^d The proportion of post-menopausal women with an assay score >25 was estimated by multiplying the proportion of post-menopausal patients from the RxPONDER study with the number of excluded patients who had an assay score >25: 1,035 * 0.669 = 692. The resulting number of patients was divided by the total amount of post-menopausal women who had a score = 0–13, 14–25, and >25: 692/(1,492 + 1,837 + 692) = 0.172.

^e The percentage of patients with different risk scores based on the MammaPrint were calculated using the number of patients who were HR+, HER2–, node positive, and had a clinically high risk in combination with a genomic low risk (687/(687 + 302) = 69.5%) and a genomic high risk (302/(687 + 302) = 30.5%), as reported in the supplementary materials of the source⁹.

^f Since all patients in the model were considered as having a clinical high risk for developing recurrences, 100% of the “no GEP testing” strategy were in the high-risk group.

^g The 5-year recurrence rate of the patients in the “no GEP testing” strategy was assumed to be the weighted average of those that were used for the Oncotype DX strategy: (0.371 * 0.073) + (0.458 * 0.087) + (0.171*0.164) = 0.095.

^h A mean body surface area of 1.75 m² was used to calculate the drug dosage for ACT.

ⁱ The treatment regimens for ACT were supported with four times G-CSF on average, which was based on the literature¹⁹ and an expert’s opinion.

Abbreviations. 3 × FEC100 + 3 × T, three cycles of fluorouracil-epirubicin-cyclophosphamide followed by three cycles of docetaxel; 4×(dd)AC + 12 × P(weekly), four cycles of doxorubicin-cyclophosphamide followed by 12 cycles of paclitaxel; 4×(dd)AC + 4 × T, four cycles of doxorubicin-cyclophosphamide followed by four cycles of docetaxel; ACT, adjuvant chemotherapy; EO, expert opinion; GEP, gene expression profiling; G-CSF, granulocyte colony stimulating factor; HER2–, human epidermal growth factor receptor 2 negative; HR+, hormone receptor positive; N1, nodal stage (1–3 positive lymph nodes).

Figure 2. Modeled results on costs per cost category and total costs for the entire population, for each strategy over a period of 5 years. The costs for GEP testing were zero for the no GEP testing strategy, because no GEP test was performed. The cost for treating recurrences was equal between the Oncotype DX and no GEP testing strategies, because the chance of developing a recurrence in the no GEP testing strategy was a weighted average of the ones that were used for the different risk groups in the Oncotype DX strategy. Abbreviations. ACT, adjuvant chemotherapy; AE, adverse event; GEP, gene expression profiling.

Figure 3. Modeled incremental results on costs per cost category and total costs for the entire population, for each strategy over a period of 5 years. The incremental costs for treating recurrences were zero between the no GEP testing and Oncotype DX strategies, because the costs for treating recurrences were equal. Abbreviations. ACT, adjuvant chemotherapy; AE, adverse event; GEP, gene expression profiling.

IKNL. NKR-cijfers. Available from: https://iknl.nl/nkr-cijfers

 Google Scholar

DICA. NBCA jaarverslag. 2020.

 Google Scholar

Kalinsky K, Barlow WE, Gralow JR, et al. 21-gene assay to inform chemotherapy benefit in node-positive breast cancer. N Engl J Med. 2021;385(25):2336–2347. doi: 10.1056/NEJMoa2108873.

 PubMed Web of Science ®Google Scholar

Piccart M, van’t Veer LJ, Poncet C, et al. 70-gene signature as an aid for treatment decisions in early breast cancer: updated results of the phase 3 randomised MINDACT trial with an exploratory analysis by age. Lancet Oncol. 2021;22(4):476–488. doi: 10.1016/S1470-2045(21)00007-3.

 PubMed Web of Science ®Google Scholar

Mamounas EP, Russell CA, Lau A, et al. Clinical relevance of the 21-gene Recurrence Score(®) assay in treatment decisions for patients with node-positive breast cancer in the genomic era. NPJ Breast Cancer. 2018;4(1):27. doi: 10.1038/s41523-018-0082-6.

 PubMedGoogle Scholar

IKNL-NABON. Borstkanker. Landelijke richtlijn. Version 2.0. 2020.

 Google Scholar

Retèl VP, Byng D, Linn SC, et al. Cost-effectiveness analysis of the 70-gene signature compared with clinical assessment in breast cancer based on a randomised controlled trial. Eur J Cancer. 2020;137:193–203. doi: 10.1016/j.ejca.2020.07.002.

 PubMed Web of Science ®Google Scholar

NZa code 050531. 2022. Available from: https://zorgproducten.nza.nl/ZoekZorgproduct.aspx?psId=15&pId=19058

 Google Scholar

NZa code 050530. 2022. Available from: https://zorgproducten.nza.nl/ZoekZorgproduct.aspx?psId=15&pId=85958

 Google Scholar

Z-Index. G-standaard April 2021. 2021. Available from: https://www.z-index.nl/

 Google Scholar

NZA. DBC-zorgproduct 020107034. 2022. Available from: https://www.opendisdata.nl/msz/zorgproduct/020107034

 Google Scholar

Thomas RJ, Williams M, Marshall C, et al. The total hospital and community UK costs of managing patients with relapsed breast cancer. Br J Cancer. 2009;100(4):598–600. doi: 10.1038/sj.bjc.6604911.

 PubMed Web of Science ®Google Scholar

Data availability statement

The input data supporting this cost-consequence model are from previously reported publications and reports, which have been cited. No datasets were generated or analyzed during the current study.