A Cost-effectiveness Analysis of Multigene Testing for All Patients With Breast Cancer

Key Points Question Is unselected genetic testing of all women with breast cancer cost-effective compared with testing based on clinical criteria or family history? Findings In this cost-effectiveness microsimulation modeling study incorporating data from 11 836 women, unselected BRCA1/BRCA2/PALB2 testing at breast cancer diagnosis was extremely cost-effective compared with BRCA1/BRCA2 testing based on clinical criteria or family history for UK and US health systems, with incremental cost-effectiveness ratios of £10 464 or £7216 and $65 661 or $61 618 per quality-adjusted life-year, respectively. One year’s unselected panel genetic testing could prevent 2101 cases of breast or ovarian cancer and 633 deaths in the United Kingdom and 9733 cases and 2406 deaths in the United States. Meaning These findings support changing current policy to expand genetic testing to all women with breast cancer.

The average number of first or second-degree relatives, ages relative to index cases, and the probability of being female are derived from data from the Office for National Statistics (UK) 28  All costs are adjusted for 2016 price index. Costs were converted wherever needed using the Hospital and Community Health-Service-Index. 63 Costs of breast cancer (BC), ovarian cancer (OC) and excess coronary heart disease (CHD) are included. In line with NICE recommendations, future healthcare costs not associated with BC, OC, or CHD were not considered. 64

Cost of genetic testing/counselling
The cost of BRCA1/BRCA2/PALB2 testing is based on testing costs for these genes in the PROMISE research programme as well as confirmatory testing costs in an accredited national genetics laboratory for those testing positive. The UK national unit cost assumed for genetic counselling is £44 per hour of client contact from PSSRU Unit costs of Health and Social Care 2010. 34

RRSO costs
The Reconstruction rate after contralateral prophylactic mastectomy (CPM) is 90%. 52 Complication rates for contralateral mastectomy are higher than unilateral mastectomy and the major complication rate with reconstruction is higher than without reconstruction. The complication rate for contralateral mastectomy without reconstruction is 42.9% (40.9% minor and 2% major) 52

Costs of breast cancer
In the general population, 10% breast cancer is non-invasive DCIS and 90% is invasive. 95% of invasive breast cancer is early and locally advanced (stage 1-3), and 5% of invasive breast cancer is advanced breast cancer (stage 4). 55 In BRCA1/2 carriers, 20% of cancers are DCIS and 80% invasive. 66,67 Stage distribution in PALB2 carriers is assumed to be the same as in the general population, owing to a lack of robust PALB2 specific data.
Annual breast cancer treatment costs in the USA are obtained from Grann et al 2011, 40  Breast surgery costs include costs of breast conserving surgery (assumed for all non-invasive cancers, and 75% of early/locally advanced invasive cancers) and costs of mastectomy (for 25% early/locally advanced and all advanced cancers). Reconstruction rates following mastectomy are reported to be 34% in the UK 77 and 55% in the US. 52 The complication rate following mastectomy alone is 21.5% (19.5% minor and 2% major) 52 and complication rate following mastectomy and reconstruction is 28.6% (24.5% minor & 4.1% major). 52 Costs are obtained from the national NHS reference costs 38 .
Chemotherapy and radiotherapy costs: Invasive breast cancers who are not at low risk 75 Recurrence costs: For non-invasive breast cancers, the non-invasive and invasive relapse rates are both 12.5%. 35% of early and locally advanced invasive breast cancers progress to advanced disease 55 . The recurrence rates for early and locally advanced breast cancer are 15.9% for node-positive 83 and 11% for node-negative disease 84 . Weighted for 31% node positive and 69% node negative, the composite recurrence rate for early and locally advanced breast cancer is 12.5%. The recurrence rate for the advanced disease is 66% (34% relapse-free five-year survival) 85 .  39 and Grann 2011. 40 16.3% uptake is assumed for chemoprevention. 25

Cost of excess CHD: British Heart Foundation statistics reports costs per capita across four
Commissioning Regions in England (London, Midlands and East, North and South) 59 .
The costs of CHD and stroke are averaged across the four regions. The prevalence of CHD is estimated at 12.0% in the UK 59 and 11.7% in the USA 60 with the onset of CHD estimated at 55 years of age. 26,58 The yearly cost of CHD in the UK is obtained by dividing the per capita cost by the population prevalence of CHD. 59 Using the report published by the American Heart Association, 61

eMethods 1. Examination of Productivity Loss
The retirement ages for females are 65 in the UK and 62 in the USA. The female labour force participation rates are 56.77% in the UK and 55.99% in the USA, obtained from the World Bank 88 .
The hourly wage rage are obtained from Office for National Statistics UK 89 and Bureau of Labour Statistics USA 90 .
We categorised the productivity costs as three subcomponents: 1) temporary disability due to shortterm work absences following diagnosis, 2) permanent disability due to reduced working hours following a return to work or workforce departure; and 3) premature mortality due to death before retirement 91 , detailed below. Reduced hours per week after returning to work (hours) 5.5 5.5

(3) Premature mortality (before retirement)
Percentage of permanent disability: workforce departure 12.9% 30% 3 The descriptive statistics for productivity loss in breast cancer patients are obtained from Hanly et al.
2012 91 . 1 We assume 98% ovarian cancer patients have cancer-related short-term work absences after diagnosis. 2 We assume ovarian cancer patients experience four weeks for surgery, 24 weeks for chemotherapy, and 24 weeks for recurrence treatment with the recurrence rate of 80% 92 . 3 We assume the percentages of permanent disability for ovarian cancer are 40% for reduced working hours and 30% for workforce departure.
We estimated temporary disability as time absent from work multiplied by age-specific gross earnings.
We calculated productivity costs due to permanent disability by applying age-specific gross earnings to the reduction in working hours, or the number of working hours if permanent workforce departure, until retirement age. Regarding productivity loss from premature mortality, we assumed that without cancer, the productive capacity of an individual would continue from the age of diagnosis until age of retirement. We multiplied the projected years of life lost by the age-specific gross earnings for the remainder of the working life to generate monetary estimates.

eMethods 2. Estimates for Age of Onset and Survival for Breast and Ovarian Cancers
Our analysis incorporates lifetime risks and long-term consequences providing a lifetime time- BC and OC survival (from diagnosis to death) were modelled using ten-year survival-data. After tenyears, we assumed the probability of death for all patients was same as the general-population. The excess risk of CHD following premenopausal oophorectomy is incorporated in the analysis. 26,99 We incorporated the fact that contralateral BC is associated with a higher risk of dying from BC. 100 We assume no significant long-term survival difference between germline and sporadic breast/ovarian cancers. 94

eMethods 4. Patient and Public Involvement Statement
The study team has worked closely with patient support groups like BRCA Umbrella and Ask Eve.
Increasing access to genetic-testing at cancer diagnosis has been highlighted to the team as an important issue affecting women with cancer and BRCA-carriers. Patients have indicated the need for access to unselected genetic-testing for BC. This has been highlighted at patient support days organised and attended by team members as well as in personal communication with leading patient stakeholders (e.g. Caroline Presho, BRCA-Umbrella). This work justifies relaxing testing guidelines, a key need highlighted by patients'. Patients did not directly input into the design and conduct of this analysis.
Patient support groups and charities will be involved in dissemination of these research findings following acceptance for publication.