Health Benefits and Cost-Effectiveness of Children's Oncology Group Breast Cancer Screening Guidelines for Chest-Irradiated Hodgkin Lymphoma Survivors

PURPOSE

To evaluate the outcomes and cost-effectiveness of the Children's Oncology Group Guideline recommendation for breast cancer (BC) screening using mammography (MAM) and breast magnetic resonance imaging (MRI) in female chest-irradiated childhood Hodgkin lymphoma (HL) survivors. Digital breast tomosynthesis (DBT), increasingly replacing MAM in practice, was also examined.

METHODS

Life years (LYs), quality-adjusted LYs (QALYs), BC mortality, health care costs, and false-positive screen frequencies of undergoing annual MAM, DBT, MRI, MAM + MRI, and DBT + MRI from age 25 to 74 years were estimated by microsimulation. BC risks and non-BC mortality were estimated from female 5-year survivors of HL in the Childhood Cancer Survivor Study and the US population. Test performance of MAM and MRI was synthesized from HL studies, and that of DBT from the general population. Costs (2017 US dollars [USD]) and utility weights were obtained from the medical literature. Incremental cost-effectiveness ratios (ICERs) were calculated.

RESULTS

With 100% screening adherence, annual BC screening extended LYs by 0.34-0.46 years over no screening. If the willingness-to-pay threshold to gain a quality-adjusted LY was ICER < $100,000 USD, annual MAM at age 25-74 years was the only cost-effective strategy. When nonadherence was taken into consideration, only annual MAM at age 30-74 years (ICER = $56,972 USD) was cost-effective. Supplementing annual MAM with MRI costing $545 USD was not cost-effective under either adherence condition. If MRI costs were reduced to $300 USD, adding MRI to annual MAM at age 30-74 years could become more cost-effective, particularly in the reduced adherence condition (ICER = $133,682 USD).

CONCLUSION

Annual BC screening using MAM at age 30-74 years is effective and cost-effective in female chest-irradiated HL survivors. Although annual adjunct MRI is not cost-effective at $545 USD cost, it could become cost-effective as MRI cost is reduced, a plausible scenario with the emergent use of abbreviated MRI.

INTRODUCTION

Chest-irradiated female childhood cancer survivors are at high risk for subsequent breast cancer (BC)¹; more than 80% of these BCs develop in Hodgkin lymphoma (HL) survivors.² The Children's Oncology Group (COG) Long-term Follow-up Guidelines recommend annual mammography (MAM) and breast magnetic resonance imaging (MRI) starting at age 25 or 8 years after chest radiation, whichever is later.³ Screening discontinuation age is unspecified.

CONTEXT

• Key Objective

• Childhood Hodgkin lymphoma survivors treated with ≥ 10 Gy of chest radiation are at elevated risk of subsequent breast cancer and are recommended to undergo annual mammography (MAM) with adjunct breast magnetic resonance imaging starting at varying screening ages by the Children's Oncology Group Guidelines, but the benefits and cost-effectiveness of these recommendations are uncertain.

• Knowledge Generated

• Using a microsimulation model, we showed that the most cost-effective strategy at the commonly cited cost-effectiveness thresholds is annual MAM from age 30 to 74 years. Adjunct breast magnetic resonance imaging is not cost-effective at the current costs, but could become cost-effective if the costs were to be reduced to $300 US dollars.

• Relevance (B.G. Haffty)

• Promoting regular screening using MAM for chest-irradiated Hodgkin lymphoma survivors from age 30 years will save lives and is cost-effective. Guidelines for survivors of childhood cancers should take into consideration cost-effectiveness in their screening recommendations.*

  *Relevance section written by JCO Deputy Editor Bruce G. Haffty, MD.

Adding MRI to annual MAM is cost-effective in BRCA1 mutation carriers.^4,5 However, high competing therapy-related non-BC mortality risks in childhood cancer survivors^6,7 may abrogate BC screening benefits. BC risks vary with chest radiation dose and other HL treatments.^8-10 Treatments have evolved over time.^11-14 Their effects on the benefits and cost-effectiveness of COG recommendations are unknown. Digital breast tomosynthesis (DBT), becoming widely adopted and considered more sensitive for dense breasts,^15,16 is potentially advantageous in younger chest-irradiated HL survivors. Since a randomized clinical trial is not feasible because of the relatively small number of childhood HL survivors¹⁷ and the long follow-up required, we used discrete-event microsimulation leveraging risk estimates from a large cohort of childhood HL survivors to compare annual MAM, MRI, DBT, MAM + MRI and DBT + MRI in base case and 25 additional BC screening strategies.

METHODS

For base case strategies, we initially compared three screening age intervals: (1) age 25 years→life, (2) age 25 years→74 years,^18,19 and (3) age 25 years or 8 years after chest radiation (whichever is later)→age 74 years. Using the selected interval, we calculated life years (LYs), quality-adjusted LYs (QALYs), BC mortality, false-positive screen rates, and health care costs by strategy, including no screening. Cost-effectiveness was assessed using the incremental cost-effectiveness ratio (ICER) or the additional cost to gain a QALY of a strategy compared with the next most effective strategy.²⁰ Multiple strategies were compared by eliminating unfavorable strategies with higher cost yielding lower QALYs than others or those whose cost-effectiveness compared with the next best alternative was lower than that of more effective strategies. The ICERs of remaining potentially cost-effective strategies were computed successively with the next effective, less costly strategy.²¹ The study was conducted from the payer and societal perspectives.²¹

Microsimulation Model

We adapted a model previously developed for the cost-effectiveness study of adjunct MRI to MAM in BRCA1/2 mutation carriers.^5,22,23 In each 3-month cycle, a woman remained healthy, developed BC, or died of non-BC. BC development included ductal carcinoma in situ (DCIS), a nonobligate precursor of invasive BC with some lesions that stabilize, but do not regress. Tumor growth followed a logistic model with log-normally distributed growth rate (mean µ and standard deviation σ) from which individuals' growth rates were drawn at the BC onset. The model comprised 11 unobservable parameters: µ, σ, probability of BC being invasive at onset, probability of DCIS being nonpalpable, annual DCIS stabilization rate, annual rate of undiagnosed DCIS becoming invasive, and five adjustment factors for BC incidence and non-BC mortality. They were estimated by systematically adjusting the estimates until simulation outputs closely matched four observable parameter values: proportion of invasive BC < 2 cm (0.339), proportion of DCIS among BC (0.046),^5,22 cumulative BC incidence (38.8%), and non-BC mortality (29.8%) at age 53 years in HL survivors in the Childhood Cancer Survivors Study (CCSS; Data Supplement, online only).

Annual BC incidence and non-BC mortality were estimated as functions of age and treatment exposures from 1,057 5-year female HL survivors in the CCSS diagnosed during 1970-1999, exposed to ≥ 10 Gy of chest radiation, and followed until December 31, 2013 (Data Supplement). Because of limited follow-up, risks beyond age 53 years were extrapolated by adding excess risk at that age to population BC risk and non-BC mortality.²⁴ The CCSS was approved by the institutional review boards at the 26 participating centers. Participants provided informed consent. The current study was determined to not involve human participants (City of Hope, Duarte, CA).

Five-million women were sampled randomly with replacement from among 1,037 female chest-irradiated CCSS HL survivors alive at age 25 years. They were assumed to be BC-free and underwent screening with 100% adherence. Age-specific BC incidence and non-BC mortality were calculated applying individuals' treatment exposures. Those with screen-positive results underwent diagnostic workup to establish diagnosis. True-positive women received BC treatment appropriate for their disease stage and estrogen receptor (ER) status. BC mortality risks of women treated for sporadic BC, given BC stage, ER status, and age at diagnosis, were used.¹ Those with false-positive and true-negative results returned to the BC-free state and continued screening. Those with false-negative results continued to progress until the next screen or clinical presentation of BC. A lifetime horizon was used.

MAM test performance was synthesized from four studies of chest-irradiated HL survivors,^25-28 adjusted for age (< 50 years, ≥ 50 years)²⁹ and BC invasiveness, using detection thresholds from previous studies^22,23 (Table 1). MRI sensitivity was synthesized from HL studies, but because their specificities were high,^31,32 we used the median of MRI specificities from 10 cost-effectiveness studies of BRCA1/2 mutation carriers.^{4,5,22,23,33-36,39,40} Lacking HL-specific estimates, test performance of DBT was synthesized (Data Supplement), and that for diagnostic ultrasound and MAM was obtained from general clinical population.^{30,37,38,41,42} Test performances of MAM + MRI and DBT + MRI were derived assuming test independence.⁴³

TABLE 1.

Test Performance Characteristics of Breast Cancer Screening and Diagnostic Modalities

Probabilities of biopsy, BC metastasis, positive lymph node by tumor size, ER status, and BC mortality by stage and age at diagnosis were obtained from the published literature.^22,44 Utility weights for healthy women⁴⁵ were adjusted by disutility in adult female childhood cancer survivors for the non-BC health state⁴⁶ and further reduced by BC stages (Data Supplement).⁴⁵ We obtained costs of screening and related procedures from 2017 Centers for Medicare & Medicaid Services,⁴⁷ and costs of BC treatment, general health care, and hospitalization costs before non-BC death from the literature (Table 2).^49-51,53 Patient costs are given in the Data Supplement. Costs were adjusted to 2017 US dollars (USD; medical portion of the Consumer Price Index, US Bureau of Labor Statistics).⁵⁴ Costs and QALYs were discounted 3% annually.

TABLE 2.

Health Care Costs

Subgroup analyses were performed by HL diagnosis year, chest radiation dose, and age at radiation. For sensitivity analyses, we varied the test performance of screening modalities, BC risk, non-BC mortality, utility weights, MRI cost, and screening adherence rate and used multiplicative risk extrapolation. We examined 25 additional strategies of varying screening age intervals and frequencies. Model validity was ensured by verifying that results varied in a consistent direction with changing input parameter values.

RESULTS

HL was diagnosed in the CCSS cohort at a median age of 15.8 years. The median age at the end of follow-up was 43.6 years. The mean chest and pelvic radiation doses were 33.6 Gy and 11.7 Gy, respectively. Among the 39.6% exposed to an anthracycline, the mean total cumulative anthracycline dose was 207.7 mg/m². Among the 56.9% exposed to an alkylator, the mean total cumulative alkylator dose was 7,182.63 mg/m² (Data Supplement). Altogether, 226 HL survivors died, 83.2% from non-BC/unknown causes (Data Supplement). Overall, 202 survivors developed BC (72% invasive) at a mean age of 39.8 years. At age 53 years, the cumulative incidence of BC was 39% and the non-BC mortality was 30% (Data Supplement). BC incidence increased with increasing chest radiation and anthracycline doses and decreased with increasing pelvic radiation and alkylator doses. Non-BC mortality increased with chest and pelvic radiation and anthracycline doses (Data Supplement).

Screening Age

Absolute differences in LY (≤ 0.017) and BC mortality (≤ 0.06 percentage point [pp]) between lifetime screening and terminating after age 74 years were small across strategies. Absolute differences between initiating screening at age 25 years for all versus age 25 years or 8 years after chest radiation, whichever is later, and terminating after age 74 years were also small (LY ≤ 0.007; BC mortality ≤ 0.03 pp) (Table 3). We therefore applied 25-74 years as the screening age interval for subsequent analyses.

TABLE 3.

LYs and Breast Cancer Mortality Risks by Screening Age Intervals

Outcomes

Without screening, mean LYs and QALYs were 64.9 and 16.6, respectively; the lifetime BC risk was 42.7%, and the BC mortality was 18.1% (Table 4, 100% adherence). The average BC risk and non-BC mortality at age 53 years relative to the general population were RR_BC53 = 7.3 and RR_non-BCmort53 = 5.1, respectively. Compared with no screening, LY and QALY gains from annual screening were 0.34-0.46 and 0.073-0.102, respectively, and the BC mortality reduction was 1.21-1.76 pp. Adding MRI to annual MAM or DBT further increased the LY by 0.11-0.12, the QALY by 0.026-0.027, and reduced BC death by 0.45-0.48 pp.

TABLE 4.

Health and Economic Outcomes by Annual Imaging-Based Screening Strategies Applied at Age 25-74 Years

Cost-Effectiveness

Here, we describe results from the payer perspective; ICERs in the societal perspective were less favorable (Table 4). Compared with no screening, costs per additional QALY gain ranged from $58,726 USD (annual MAM) to $155,340 USD (annual DBT + MRI). In multiple comparison, the strategies that remained after eliminating unfavorable strategies (see the Methods section) were annual MAM at age 25-74 years (ICER = $58,726 USD, v no screening) and annual MAM + MRI at age 25-74 years (ICER = $385,285 USD, v annual MAM age 25-74 years). If the threshold for willingness-to-pay (WTP) to gain a QALY was ICER < $100,000 USD, only annual MAM could be considered cost-effective. Adding MRI to annual MAM was not cost-effective unless the WTP threshold was < $400,000 USD.

Across strategies, reducing 100% screening adherence to moderate adherence⁵⁵ (65% of survivors with 100% adherence, 25% with 50% adherence, and 10% with 0% adherence) reduced the LY and QALY by 13%-20% and costs by 0.6%-2% and increased the ICER by 3%-11%. Cost-effectiveness results remained consistent (Table 4).

Subgroup Analysis

As chest radiation dose increased, pelvic radiation dose increased, anthracycline and alkylator doses decreased, and RR_BC53 and RR_non-BCmort53 increased (Data Supplement). Without screening, LYs were highest (76.3) for the lowest chest radiation dose and lowest (59.4) for the highest chest radiation dose groups (Data Supplement). With screening, LY and QALY gains were 0.31-0.62 and 0.07-0.15, respectively, across strategies and radiation dose groups. BC death declined by 1.1-2.1 pp. Across chest radiation doses, only annual MAM was cost-effective at the WTP threshold < $100,000 USD. The ICERs of adding MRI to annual MAM or DBT remained at > $250,000 USD.

From 1970-1974 to 1995-1999, mean chest radiation dose (38.6 Gy→25.4 Gy), mean pelvic radiation dose (16.4 Gy→4.0 Gy), and RR_BC53 and RR_non-BCmort53 declined, whereas the percentage exposed to anthracyclines increased (7.2%→94.4%; Data Supplement). Without screening, LYs (57.8→72.8) and BC mortality (15.1%→20.3%) increased with era (Data Supplement). Across eras, only annual MAM was cost-effective at the WTP threshold < $100,000 USD. The ICERs of adding MRI to annual MAM or annual DBT across eras were > $340,000 USD.

For survivors exposed to chest radiation at age < 12 years or ≥ 12 years, only annual MAM was cost-effective at WTP threshold < $100,000 USD. ICERs of adding MRI to annual MAM or DBT were > $300,000 USD in both groups (Data Supplement).

Sensitivity Analyses

Sensitivity analyses for the cohort (Data Supplement) showed that the ICER of adding MRI to annual MAM remained at > $190,000 USD per QALY across multiple scenarios: (A) increased MRI sensitivity for DCIS (0.50→0.72; ICER = $ 346,226 USD); (B) increased MRI specificity (0.81→0.93; ICER = $365,931 USD); (C) both (A) and (B) (ICER = $328,826 USD); (D) removed utility decrement for childhood cancer survivors (ICER = $364,329 USD); (E) BC incidence increased by 50% (ICER = $231,717 USD); (F) BC incidence decreased by 50% (ICER = $776,826 USD); (G) non-BC mortality increased by 50% (ICER = $398,098 USD); (H) non-BC mortality decreased by 50% (ICER = $295,738 USD); (I) BC-specific mortality increased by 30% (ICER = $345,268 USD); (J) risk extrapolation used multiplicative assumption (ICER = $352,173 USD); reduced MRI cost to $300 USD (K.1) assuming 100% adherence (ICER = $223,999 USD) or (K.2) moderate adherence (ICER = $220,925 USD), (K.3) for chest radiation at age < 12 years (ICER = $343,980 USD) or (K.4) ≥ 12 years (ICER = $193,290 USD); and (L) used alternative MAM and MRI test performance estimated from a more recent larger chest-irradiated childhood cancer survivor cohort⁵⁶ (ICER = $226,392 USD).

Additional sensitivity analyses showed that ICER of adjunct MRI was sensitive to values of MAM sensitivity and MRI cost. Adding MRI to annual MAM assuming a lower MAM sensitivity of BRCA1/2 mutation carriers (40%)⁵⁷ increased the sensitivity of MAM + MRI proportionally more than assuming MAM sensitivity of HL survivors (70%), resulting in a greater LY gain (0.31 years v 0.11 years), BC mortality reduction (1.2 v 0.5 pp), and better cost-effectiveness of adjunct MRI (ICER $143,022 USD v $385,285 USD [base case]). Cost-effectiveness improved further (ICER < $100,000 USD) when the MRI cost was reduced to $300 USD (Data Supplement).

Screening Outcomes

The number of false positives was higher in strategies involving MRI than MAM-only strategies (MAM + MRI = 294.3, DBT + MRI = 239.2, MRI = 211.1, MAM = 68.8, and DBT = 22.6 per 1,000 screens). Because of lower MRI versus MAM specificity, > 50% of survivors had ≥ 4 lifetime false-positive results undergoing annual MRI, MAM + MRI, or DBT + MRI compared with < 15% for annual MAM and annual DBT. One BC death was averted for every ≥ 300 false-positive screens for MRI, MAM + MRI, and DBT + MRI compared with < 150 for MAM or DBT (Table 5).

TABLE 5.

Screening Test Results From Five Million Simulated HL Survivors

Other Strategies

To identify potentially more cost-effective BC screening strategies, we examined 30 strategies of varying screening age intervals and frequencies, at the MRI cost of $300 USD and $545 USD, assuming moderate and 100% adherence (Data Supplement). After eliminating unfavorable strategies (see the Methods section), the remaining strategies were graphed by cost and QALY and connected by a line on a cost-effectiveness frontier, below which are shown the eliminated strategies (Fig 1). The ICERs of successive strategies on the frontier were calculated compared with the next most effective strategy. At the MRI cost of $545 USD, annual MAM at age 25-74 years (ICER = $74,323 USD, 100% adherence) or annual MAM at age 30-74 years (ICER = $56,372 USD, moderate adherence) would be considered cost-effective at the WTP threshold < $100,000 USD. A WTP threshold > $250,000 USD is needed to consider supplemental MRI as cost-effective. If the MRI cost was reduced to $300 USD, supplemental MRI could be cost-effective at the WTP threshold < $160,000 USD (annual MAM at age 25-74 years with biennial MRI at age 30-64 years, ICER = $157,520 USD) assuming 100% adherence or at the WTP threshold < $140,000 USD (annual MAM + MRI at age 30-74 years, ICER = $133,682 USD) assuming moderate adherence. Annual MAM + MRI at age 25-74 years was too costly (ICER > $400,000 USD) or was eliminated because there were better alternatives.

FIG 1.

Cost-effectiveness frontier: strategies involving digital MAM, breast MRI, and DBT from the payer perspective: (A) 100% adherence and (B) moderate adherence. Strategies below the cost-effectiveness frontier are less effective and more costly than at least one other strategy (dominated). Details of the strategies are given as follows: 0, no screening; 1, annual MAM age 40-74 years; 2, biennial (MAM + MRI age 25-74 years); 3, annual MAM age 30-74 years; 4, annual (MRI age 25-39 years/MAM age 40-74 years); 5, annual DBT age 25-74 years; 6, annual (MRI age 25-49 years/MAM age 50-74 years); 7, annual MAM age 25-74 years; 8, annual (MRI age 25-59 years/MAM age 60-74 years); 9, annual MRI age 25-74 years; 10, annual MAM + quinquennial MRI age 25-74 years; 11: annual MAM age 25-74 years/biennial MRI age 25-39 years; 12, annual MAM age 25-74 years/triennial MRI age 25-54 years; 13, annual MAM + quadrennial MRI age 25-74 years; 14, annual MAM age 25-74 years/triennial MRI age 25-49 years; 15, annual MAM age 25-74 years/triennial MRI age 25-64 years; 16, annual ([MAM + MRI] age 25-39 years/MAM age 40-74 years); 17, annual MAM + biennial MRI age 25-74 years; 18, annual MAM age 25-74 years/biennial MRI age 25-49 years; 19, annual MAM age 25-74 years/triennial MRI age 25-74 years; 20, annual MAM age 25-74 years/biennial MRI age 25-54 years; 21, annual MAM age 25-74 years/biennial MRI age 30-64 years; 22, annual MAM age 25-74 years/biennial MRI age 25-64 years; 23, annual ([DBT + MRI] age 25-39 years/DBT age 40-74 years); 24, annual ([MAM + MRI] age 25-49 years/MAM age 50-74 years); 25, annual (MAM + MRI) age 30-74 years; 26, annual (DBT + MRI age 25-49 years/DBT age 50-74 years); 27, annual ([MAM + MRI] age 25-59 years/MAM age 60-74 years); 28, annual ([DBT + MRI] age 25-59 years/DBT age 60-74 years); 29, annual (DBT + MRI) age 25-74 years; and 30, annual (MAM + MRI) age 25-74 years. ICER is the cost per QALY gained compared with another strategy; Cost and QALY were discounted by 3% annually. DBT, digital breast tomosynthesis; ICER, incremental cost-effectiveness ratio; K, thousand; MAM, mammography; MRI, magnetic resonance imaging; QALY, quality-adjusted life year; USD, US dollars.

DISCUSSION

Our comparative modeling analysis of BC screening strategies for female survivors of childhood HL showed that initiating screening at age 25 years for all, instead of 8 years after chest radiation for some, and terminating after age 74 years do not compromise long-term health outcomes, suggesting that the current COG recommendations may be simplified. At commonly cited WTP thresholds < $100,000 USD or < $150,000 USD,⁵⁸ the most cost-effective strategy examined in the base case assuming 100% screening adherence was annual MAM at age 25-74 years, providing 0.35 LY and 0.075 QALY over no screening (ICER = $58,726 USD). Supplementing this strategy with annual MRI further added 0.11 years and 0.027 QALY, but at a cost well exceeding the above WTP thresholds. Cost-effectiveness results were consistent across chest radiation dose, HL diagnosis eras, age at chest radiation, varying screening adherence rates, and robust across most sensitivity analysis scenarios.

When 30 strategies of varying screening age intervals and frequencies were examined at the MRI cost of $545 USD, annual MAM at age 30-74 years was most cost-effective at the WTP threshold < $60,000 USD for either adherence. In the more realistic situation of moderate adherence, starting MRI earlier at age 25 years would not be considered cost-effective even after raising the WTP threshold to $150,000 USD, nor would supplementing the MAM-only strategy with MRI be cost-effective. However, adjunct MRI could become cost-effective if MRI costs were reduced to $300 USD, particularly under moderate adherence, where the cost to gain a QALY of adding MRI to annual MAM at age 30-74 years is approximately $134,000 USD. Hence, from the cost-effectiveness standpoint, our preference is annual MAM at age 30-74 years, with adjunct MRI if its cost decreases. This is plausible with abbreviated MRI.^59-61

ICER of adjunct MRI was affected by MAM sensitivity. Adding MRI to MAM already at high sensitivity (70%) did not substantially improve BC detection. Coupled with high non-BC mortality in HL, BC-specific survival benefit of BC screening was abrogated. Adjunct MRI was more cost-effective (ICER < $180,000 USD) when MAM sensitivity was lower (40%). Cost-effectiveness improved further (ICER < $103,000 USD) when the MRI cost was reduced to $300 USD. Whether the MAM sensitivity of 40% holds in HL is questionable. The four HL studies used to synthesize the test performance, albeit small, showed consistently high sensitivity for MAM (66.7%-70%),^25-28 which was attributed to radiographic and pathologic features and natural history of the BC in HL.²⁷ A similar result of ICER = $226,392 USD for adjunct MRI was obtained using test performance estimates from a larger study of female chest-irradiated childhood cancer survivors.⁵⁶ As noted, adjunct MRI could become more cost-effective by reducing MRI cost, a plausible scenario with emergent use of abbreviated MRI whose test performance is comparable with full-protocol breast MRI but with lower costs.^59-61

A Canadian study of chest-irradiated adolescent HL survivors assuming moderate adherence found annual MAM at age 25-74 years as cost-effective at the WTP threshold < $50,000 Canadian dollars (CAD).⁵⁵ At the WTP threshold < $150,000 CAD, annual MRI at age 25-49 years and then switching to annual MAM at age 50-74 years were cost-effective. From their published results, ICER for MRI added to annual MAM at age 25-74 years could be calculated as $247,810 CAD, which is consistent with our results of ICER > $200,000 USD. Another study in chest-irradiated childhood cancer survivors assuming 100% adherence using two different BC microsimulation models compared annual MRI alone with annual MAM + MRI from age 25 or 30 years to age 74 years and showed different results by the microsimulation model; their preference at the WTP threshold < $100,000 USD was annual MAM + MRI at age 30-74 years.⁶² The annual MAM-only strategy was not examined, and the cost-effectiveness of adjunct MRI was not evaluated. Aside from the MAM sensitivity used in these studies (40%⁶² v 70%⁵⁵), the ICER of a strategy depends on alternatives considered, which partially explains the differences among studies. Nevertheless, it is notable that our report and the Canadian study found MRI adjunct to annual MAM at age 25-74 years as not cost-effective at the commonly cited cost-effectiveness thresholds.

Unlike European studies,^63-65 DBT and MAM sensitivities were comparable in average BC risk women in the United States, but with higher DBT specificity.^15,30,66 Annual DBT from age 40 years was not cost-effective at the WTP threshold < $150,000 USD in these women.⁶⁶ We found similar results for HL. Nevertheless, DBT is being adopted broadly in the US clinical practice.⁶⁷ Its higher specificity is considered advantageous in reducing recall rates in BC screening.

Strengths of our study include the use of treatment exposure-specific BC incidence and non-BC mortality risks estimated directly from the large CCSS chest-irradiated female HL cohort. Risks were applied to clinically relevant populations simulated from this cohort, enhancing realism and enabling subgroup analyses. Our BC development model incorporated detection and treatment of DCIS. A broad range of BC screening strategies were examined, under 100% and reduced adherence. Limitations include extrapolation of risks beyond age 53 years and lack of availability of HL-specific DBT test performance. As the BC screening rate in this population is low (55%),⁶⁸ an effective intervention to promote BC screening is needed.⁶⁹ We found annual MAM to be both effective and cost-effective. Although annual adjunct MRI as recommended by the COG Guidelines is not cost-effective, the decreasing MRI cost could potentially make adjunct MRI more cost-effective in chest-irradiated HL survivors.

AUTHOR CONTRIBUTIONS

Conception and design: F. Lennie Wong, Janie M. Lee, Wendy M. Leisenring, Kevin C. Oeffinger, Tara O. Henderson, Gregory T. Armstrong, Smita Bhatia

Financial support: Gregory T. Armstrong

Administrative support: Joseph P. Neglia, Gregory T. Armstrong, Smita Bhatia

Provision of study materials or patients: Joseph P. Neglia, Gregory T. Armstrong, Leslie L. Robison

Collection and assembly of data: F. Lennie Wong, Wendy M. Leisenring, Joseph P. Neglia, Rebecca M. Howell, Susan A. Smith, Gregory T. Armstrong, Leslie L. Robison

Data analysis and interpretation: F. Lennie Wong, Janie M. Lee, Joseph P. Neglia, Rebecca M. Howell, Chaya S. Moskowitz, Tara O. Henderson, Ann Mertens, Paul C. Nathan, Yutaka Yasui, Wendy Landier, Smita Bhatia

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Health Benefits and Cost-Effectiveness of Children's Oncology Group Breast Cancer Screening Guidelines for Chest-Irradiated Hodgkin Lymphoma Survivors

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/jco/authors/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).