Cost Effectiveness of Whole Population BRCA Genetic Screening for Cancer Prevention in Israel

Current recommendations for BRCA testing in Israel, as in most countries, are clinical and based upon personal or family history of breast or ovarian cancer. Using this clinical, family-history based screening strategy has the potential of missing most carriers (1, 2). With the advances in next-generation genetic sequencing and the reduction in genetic testing costs, whole population screening has become a feasible option (3). Previous studies have shown that population-based BRCA screening among Ashkenazi Jewish women which have a high prevalence of BRCA carriers is cost effective in the United States and the United Kingdom (4), also for varying degrees of Ashkenazi Jewish ancestry (5). In fact, when all four grandparents are Ashkenazi, whole population screening was found to be cost saving (4). Among Jews of Sephardic ancestry, where BRCA prevalence is lower, population-based BRCA testing was also found to be highly cost effective compared with clinical criteria approach (6). Due to several generations of mixed Ashkenazi and Sephardic marriages, the exact prevalence of BRCA carrier state in Israel is still unknown but reaches as much as 2.5% of Jews of Ashkenazi ancestry and at least 0.7% of the pure Sephardic population (7–9) perhaps making whole population screening particularly effective. The aim of our study was to investigate the cost effectiveness of whole population BRCA screening in Israel, for varying degrees of BRCA prevalence.

Lifetime costs of whole female population screening for BRCA mutation carrier state starting at the age of 30 versus nonscreening were compared using a Markovian process decision analysis model (Fig. 1). The model assumed that all women in the screening arm would have genetic screening. Women found positive for BRCA would be offered risk reduction salpingo-oophorectomy (RRSO) at age 40 for ovarian cancer prevention and breast screening with yearly breast MRI/ultrasound and or risk reduction mastectomy (RRM) for breast cancer prevention. Costs were discounted at 3%. Quality-adjusted life years (QALY), which reflect both quality and quantity of life lived and incremental cost-effectiveness ratios (ICER), were used for cost-effectiveness analysis. One-way sensitivity analysis was conducted with all variables to evaluate model uncertainty. In addition, we conducted a probabilistic sensitivity analysis (Monte Carlo simulation) using 100 trials, with each including 10,000 subjects.

BRCA screening in the model was done by using a 14-mutation panel most commonly used for BRCA genetic testing in Israel. This panel includes the three most common BRCA founder mutations [BRCA1 185delAG NM_007294.4:c.66_67AG (p.Glu23fs), BRCA1 5382insC NM_007294.4:c.5266dup (p.Gln1756fs), and BRCA2 6174delT NM_000059.4:c.5946del (p.Ser1982fs)] that account for the vast majority of inherited cancer risk due to BRCA1 and BRCA2 in Israel (8).

Screening and other treatment-related costs were received from the Israeli Ministry of Health's 2020 pricing list according to specified codes with conservative assumptions of health resources utilities (Table 1). Elaborate costs used in the model are shown in Supplementary Tables S1 to S3 for ovarian cancer treatment, breast cancer treatment, and BRCA-positive patients' follow-up, respectively.

Model probabilities are presented in Table 2. The probability of being at the end of each arm of the model constructed was calculated by multiplying the probabilities of events along the arm path. Mean age used in the model for developing sporadic ovarian cancer was 55 years (10) and 62 years for breast cancer (11). Among BRCA carriers, average ages used were 45 and 50 for ovarian and breast cancer, respectively (12, 13). The mortality rates of subjects who developed breast cancer or ovarian cancer were based on stage distribution and Kaplan–Meier survival curves for each stage, based on the Surveillance, Epidemiology and End Results program database. Survival curves were extrapolated by fitting Weibull distribution using the Nelder–Mead Algorithm.

As the true prevalence of BRCA mutation carriers in Israel is not yet known due to several generations of mixed Ashkenazi and Sephardic marriages, sensitivity analysis was done to account for varying degrees of BRCA prevalence.

Values of health benefits for each strategy (screening vs. nonscreening), from payer prospective, were calculated using QALYs. QALYs are calculated by multiplying the utility value associated with a given state of health by the number of years lived in that state, where QALY of one reflects 1 year lived in perfect health and QALY of zero represents death state. The ICER was then calculated by using the formula: (average cost screening – average cost nonscreening)/(average QALY screening – average QALY nonscreening). The ICER calculated enabled to determine whether whole population screening is cost effective or not, compared with willingness to pay threshold. An intervention was defined as cost effective if the ICER per QALY is between 1 and 3 times per capita GNP [GNP in Israel is estimated at 42,160 USD, equivalent to 142,500 new Israeli Shekels (NIS), estimated during October 2020]. Interventions below 1 GNP per capita are considered very cost effective (14, 15).

As this work is a theoretical, mathematical/financial model, not involving any human or animal subjects in any form, after consultation with our local Institutional Review Board (IRB), it was exempt from the need of IRB approval.

For a BRCA prevalence of 2.5%, the known BRCA prevalence among Jews of Ashkenazi ancestry (7, 8), whole population screening is very cost effective, with an ICER of 81,493 NIS, per QALY gained (24,110 USD), data are shown in Table 3.

One-way sensitivity analysis was conducted for all variables. Results are presented on a tornado diagram (Supplementary Fig. S1). The diagram presents how model parameters influence the ICER calculated. BRCA population prevalence and discount rate (set at 3% in our model) and RRSO acceptance rate are the most important variables affecting the ICER. As shown in Fig. 2, as the prevalence of BRCA carrier state decreases, the ICER increases. For a theoretical prevalence of 1%, the ICER is 176,395 NIS per QALY gained (52,187 USD). For a BRCA prevalence of 0.75%, which is the estimated BRCA prevalence among Sephardic Jews (9), the ICER is 250,000 NIS (73,964 USD). Whole population screening strategy compared with no screening equates to 15.4 and 6.1 days in life expectancy gain for prevalence of 2.5% and 1%, respectively, at a 3% discount rate.

Figure 3 displays the acceptability curve. The acceptability curve is a visual aid for communicating the results of probabilistic sensitivity analysis in cost-effectiveness models. Acceptability curve presents the relative cost effectiveness as a function of the ICER threshold (WTP). For each WTP value, the graph uses net benefits to determine the percentage of simulation iterations in probabilistic sensitivity analysis that favor each strategy. In our study, we found that at a willingness to pay threshold of 340,000 NIS (which are equivalent to 100,000 USD), about 90% of simulation iterations in the whole population BRCA screening would be cost effective (Fig. 3).

Each year around 360 new cases of ovarian cancer and 4,850 new cases of invasive breast cancer are diagnosed in Israel, posing a tremendous medical and economic burden on the Israeli health system. As BRCA is an actionable gene that enables very effective prevention measures, identification of carriers may have far-reaching implications. With the improving ease in the technical aspect of genetic testing and the reduction in testing costs, whole population screening is becoming ever more feasible and must be taken into account from both the medical and the economic aspects. For any screening program, event prevalence has a huge impact on screening effectiveness, and as the true prevalence of BRCA in Israel, as in most countries, is unknown, we conducted a sensitivity analysis to control for this uncertainty. Our data show that screening for BRCA mutations in Israel would be cost effective across a wide range of BRCA prevalence values, starting at the highest probable prevalence of 2.5% among Jews of pure Ashkenazi origin, where screening is very cost effective, through to a theoretical prevalence of 0.75% of population, where screening was still found to be cost effective compared with per capita GNP in Israel.

Our data are comparable with data from previous reports from other countries. Manchanda and colleagues reported that population-based screening among Ashkenazi Jewish women is cost saving with a baseline-discounted ICER of -2,079£ per QALY in the United Kingdom (4). Further analysis revealed that population-based screening is cost effective also for varying levels of Ashkenazi Jewish ancestry, in the United Kingdom and the United States (5). Even among Jews of Sephardic ancestry, who usually carry one of the three most common BRCA1 mutation, with a prevalence of 0.5% to 1% in the United Kingdom and in the United States, population-based screening was found to be cost effective (6).

The current strategy of family history–based screening lacks the sensitivity required as a screening tool and has the potential of missing over 50% of carriers at risk (1, 16, 17) and with the growing ease of genetic screening may soon be an unacceptable alternative from both medical and legal standpoint. After diagnosis, the reported uptake of RRSO among BRCA carriers varies geographically and culturally, but may reach as high as 98% at designated centers (18–21). Consultation with a gynecologic oncologist seems to play an important role in the uptake of risk reduction surgery among effected patients (21). The reported uptake of RRSO in Israel varies between 49% and 78% (22, 23). Raising the rate of RRSO uptake would improve further the cost effectiveness of screening.

Even though surgical menopause has a direct negative effect on quality of life measures such as vasomotor symptoms, sexual function, bone health as well as cardiovascular health, it is important to note that after surgery, hormone therapy is not contraindicated and even encouraged and that available literature shows that breast cancer risk following RRSO in premenopausal BRCA carriers would not change with hormone replacement therapy (HRT) use (24, 25). Moreover, quality of life does not seem to differ between carriers who underwent RRSO to carriers who opt for periodic gynecologic screening and even the general population. RRSO is associated with significantly fewer breast and ovarian cancer worries and more favorable cancer risk perception (26–28). Uptake of RRM is generally lower than that RRSO among BRCA carriers (22, 23, 28), yet those who choose to have RRM show high degree of satisfaction mainly due to the reduction in fear of cancer (29).

The main strength of our work is the sensitivity analysis conducted. This analysis enabled us to estimate cost effectiveness of screening with varying degrees of BRCA prevalence values, thus overcoming the uncertainty of the true BRCA prevalence in the general Israeli population, including subpopulations with lower estimated BRCA prevalence. This model may be used in other countries where BRCA prevalence has yet been elucidated. Other strengths of our work include the inclusion of newest, highly costly drugs such as bevacizumab and PARP inhibitors in the treatment algorithm of breast and ovarian cancer. PARP inhibitors were only recently approved for the first-line treatment of patients with BRCA-positive ovarian cancer in Israel and add considerably to the costs involved with treatment of these patients. Our algorithm also used the most up-to-date probability estimates of ovarian cancer risk reduction after RRSO, 80% in our model, and not the 96% reduction previously reported (30). Moreover, even though it seems biologically plausible that RRSO may also influence the pathophysiology of breast cancer development, the protective effect of RRSO on breast cancer, which was previously estimated at 50% (30–33), is now under debate. Several newer publications that used RRSO as a time-dependent covariate did not conquer any protective effect of RRSO on breast cancer incidence (34–36). Thus, in our model, we assumed zero effect of RRSO on breast cancer. Had we used 50% as the estimated protective effect of RRSO on breast cancer incidence, the ICER calculated would be even more cost effective than the one reported (50,494 NIS per QALY gained for a BRCA prevalence of 2.5%).

Our work has several shortcomings. Our model compared whole population screening with no screening and not to family history–based screening as is currently offered. We believe that family-based screening, although recommended, is highly opportunistic, inconsistent, and greatly depends on the awareness of the primary care physician at the time cancer diagnosis and its implementation are extremely difficult to assess. In our model, we used a very conservative estimation of costs involved with breast and ovarian cancer treatment that took into account only the basic treatment needs of the "ideal" breast/ovarian cancer patient and omitted many complications involved with the surgical and medical treatment of these patients. Even with these conservative estimations, we found screening to be cost effective. Finally, implementing a population-based genetic screening strategy, although cost effective, would raise many challenges that would need to be overcome, starting from the logistics involved in screening whole population, through the challenges of handling positive tests, down to solving ethical dilemmas that would rise once this delicate, personal genetic information is known.

In conclusion, with the progress in genetic testing technologies and decreasing costs, population-based screening for BRCA mutations, as well as for other actionable genetic mutations, may become a feasible option, displacing more obsolete recommendations such as family history–based screening.

A. Sonnenblick reports grants and nonfinancial support from Novartis, personal fees and nonfinancial support from Roche, personal fees and nonfinancial support from Medison, and personal fees and nonfinancial support from Pfizer outside the submitted work. No disclosures were reported by the other authors.

N. Michaan: Conceptualization, resources, data curation, supervision, validation, investigation, methodology, writing–original draft, project administration, writing–review and editing. M. Leshno: Conceptualization, resources, data curation, software, formal analysis, supervision, validation, investigation, methodology, project administration, writing–review and editing. T. Safra: Conceptualization, resources, data curation, validation, investigation, writing–review and editing. A. Sonnenblick: Conceptualization, resources, data curation, supervision, validation, investigation, writing–review and editing. I. Laskov: Conceptualization, resources, data curation, validation, writing–original draft, writing–review and editing. D. Grisaru: Conceptualization, resources, data curation, supervision, validation, writing–original draft, project administration, writing–review and editing.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.