Patient Preferences in Neoadjuvant Therapy for HER2+ Early-Stage Breast Cancer

Abstract

Purpose. Health technology assessment (HTA) focuses on overall survival (OS) as the key clinical endpoint when making oncology treatment decisions. However, capturing robust OS estimates in early-stage breast cancer (eBC) is challenging, and patients may value other endpoints. This study assessed patient preferences for treatment attributes and endpoints in neoadjuvant therapy where mature and nonconfounded OS is rarely available. Patients and Methods. An online discrete choice experiment (DCE) was developed with health care professionals and patient advisory groups and conducted in Germany, France, Italy, and Spain with patients with HER2+ eBC. Patients were presented with 15 tasks and in each asked to choose between 2 hypothetical treatment options or to opt out in the neoadjuvant setting. Treatment attributes included OS at 5-years, disease-free survival (DFS) at 5-years, pathological complete response (pCR), impact of side effects on quality of life (QoL), and ability to receive breast-conserving surgery (BCS). Data were analyzed using multinomial logit and random parameters logit models. Results. Three hundred thirty-four patients with HER2+ eBC responded. The most valued attribute was achieving pCR (no invasive cancer in the breast and lymph nodes in all patients after treatment), followed by 5-y DFS in 95% of patients. OS ranked third in importance. The ability to undergo BCS and the impact of side effects on QoL were less important. Preferences varied by hormone receptor status, time since diagnosis, cancer stage, and age. Conclusion. Our findings are in-line with regulatory reviews, which accept pCR, DFS, and OS as clinically valid endpoints. When recommending reimbursements for treatments, HTA bodies and payers should consider patient preferences for treatment attributes and endpoints in their decision making when valuing new cancer treatments.

Highlights

Traditional cancer treatment assessments focus on overall survival (OS), but in early-stage breast (eBC) cancer, patients may prioritize other treatment outcomes, since robust OS data may be unavailable.

A discrete choice experiment was conducted with 334 patients with HER2+ eBC in the neoadjuvant setting in Germany, France, Italy, and Spain.

Patients considered achieving pathological complete response (no invasive cancer in the breast and lymph nodes in all patients after treatment) to be the most important attribute when making a treatment decision, followed by disease free-survival (DFS) and OS; the ability to undergo breast-conserving surgery and the impact of side effects on quality of life were less important.

These insights highlight the need for health authorities and reimbursement bodies to consider patient-valued outcomes beyond OS in eBC treatment evaluations.

Keywords

discrete choice experiment early-stage breast cancer neoadjuvant therapy patient preferences

Breast cancer is the most common cancer diagnosed globally, and growing incidence rates have been observed in high-income countries. This increase has been linked to a surge in detected small, early-stage tumors.¹ Early-stage breast cancer (eBC) (also known as operable breast cancer) is defined as any form of cancer not spread beyond the breast tissue or axillary lymph nodes.² In eBC, 15% to 20% of carcinoma diagnoses reveal genetic mutations of cancer cells that have an overexpressed and/or amplified human epidermal growth factor receptor 2 (HER2) gene.³ HER2-positive (HER2+) tumors are one of the most aggressive breast cancer subtypes, and there is a need to improve outcomes for patients with HER2+ eBC.

Neoadjuvant therapy has become the standard of care for patients with HER2+ eBC, followed by surgery, chemotherapy, hormone-based treatment, and/or radiotherapy.^4,5 When the efficacy of a (novel) neoadjuvant eBC treatment is assessed, overall survival (OS; the time from the start of the study treatment until the patient’s death) remains the gold standard primary endpoint of clinical trials, given its unambiguous endpoint definition⁶ and the preferred endpoint for payers when appraising cancer treatments for reimbursement.⁷ However, it can take years to collect enough data for eBC treatments to demonstrate long-term efficacy and safety to regulators and payers, meaning these data are unavailable for regulatory and reimbursement decisions.^6,8 A strict regulatory and payer requirement for OS data for eBC can also delay access of new medicines for patients.⁷ OS analysis may also be less relevant to patients compared with other endpoints in an early setting, considering the longer life expectancy of patients with eBC compared with those in advanced stages.⁷

Although numerous non-OS endpoints are recognized as clinically valid endpoints for eBC and have received approval from regulatory agencies,⁹ challenges remain in reimbursement among innovative medicines for eBC in the absence of mature OS data, thereby delaying patient access.⁷ Regulator and payer decision making when considering surrogate endpoint measures of OS should be informed by patient perspectives, about which little is known.⁷ Patient perspectives can be captured using a patient preference study, which the Food and Drug Administration (FDA) guidance defines as “a statement of the relative desirability or acceptability to patients of specified alternatives or choices among outcomes or other attributes that differ among alternative health interventions.”¹⁰^(p6) The Innovative Medicines Initiative–Patient Preferences in Benefit-Risk Assessments during the Medical Product Lifecycle (IMI-PREFER) developed an expert and evidence-based framework, with input from patients, regulators, health technology assessment (HTA) bodies, payers, and other stakeholders, on how to design, conduct, and analyze patient preference studies.¹¹ In May 2021, the European Medicines Agency provided a positive qualification opinion on the IMI-PREFER framework. This is important for breast cancer treatments, as priorities between patients, providers, and payers may differ. Incorporating the patient voice in treatment decision making can lead to higher levels of care and satisfaction and improved health outcomes.^12,13

There are several key surrogate endpoint measures of OS used to assess disease progression relevant for HER2+ eBC in the neoadjuvant setting. One endpoint is disease-free survival (DFS; the time from randomization until recurrence of tumor or death from any cause), including invasive DFS. Evidence has shown a statistically significant association between DFS, including invasive DFS, and patients’ OS at both individual and trial levels for HER2+ eBC.¹⁴ Another surrogate endpoint demonstrating positive patient-level associations in the eBC neoadjuvant setting is pathological complete response (pCR; the absence of residual invasive cancer in the breast and axillary lymph nodes, including those who have ductal carcinoma in situ), for both OS and DFS.^15,16 Event-free survival (EFS; the time from randomization to the progression of the disease precluding surgery, local or distant recurrence, and death due to any cause) is also a relevant surrogate endpoint. In recent years, both the EMA and FDA have accepted pCR, DFS, and EFS as validated endpoints in high-risk eBC.⁸

An additional patient-relevant outcome important for HER2+ eBC patient treatment decisions is quality of life (QoL). Side effects of treatment and their impact on patients’ functional domains (physical, emotional, social, or role functioning) are often underestimated when understanding patients’ treatment decision making, and patients may prioritize experiencing fewer side effects or maintaining QoL over longer OS.^17,18

This study aimed to assess patient preferences for treatment attributes and endpoints with patients with HER2+ eBC in the neoadjuvant setting using a discrete choice experiment (DCE).

Methods

A DCE was included in an online survey to quantify preferences with patients with HER2+ eBC in the neoadjuvant setting in Germany, France, Italy, and Spain. A DCE elicits preferences by presenting participants with hypothetical scenarios comprising of alternatives, defined by attributes and a finite set of possible values the attribute can assume (attribute levels), simulating real-world situations.^19,20 These 4 European markets were selected as HTA decision making is driven by clinical effectiveness and acceptance of surrogate endpoints for OS is low.

Identifying Patient-Relevant Attributes

A targeted literature review (TLR) was conducted to explore the literature on patient preferences of eBC treatments. From 15 studies extracted from PubMed and Google Scholar, findings revealed DFS (n = 4/15) and pCR (n = 3/15) were non-OS endpoints that patients with eBC considered important for treatment decision making. For instance, Thill et al.²¹ found patients placed the most importance on pCR when making neoadjuvant treatment decisions for eBC. Patient perceptions of safety/tolerability and willingness to accept various side effects were also important (n = 7/15), as was the effect of side effects on patients’ QoL (n = 3/15). Other treatment factors included the need for surgery (n = 2/15), including BCS. For detailed information on the TLR, please refer to Appendix 1.

TLR findings informed the development of guides and 4 patient vignettes for qualitative discussions with health care providers (HCPs) and patient advisory group (PAG) members. The objectives were to 1) explore attributes driving preferences for eBC treatment decisions and 2) evaluate how endpoints were communicated between patients and oncologists. The 4 vignettes simulated hypothetical situations of eBC (2 adjuvant and 2 neoadjuvant patient profiles). Discussions were conducted through an unblinded advisory board discussion with 6 PAG members and 20 HCPs in double-blinded semi-structured one-on-one telephone interviews. Data were analyzed and coded thematically.

Qualitative discussions revealed OS, pCR, and DFS were important endpoints for patient and HCP eBC treatment decisions. In the neoadjuvant setting, pCR was noted as a key endpoint in eBC, given its correlation with OS.^15,16 DFS was also important, given it can show whether the treatment is stopping the cancer from returning. Both groups also identified attributes affecting patients’ daily lives as important, including the impact of side effects on QoL and impact of BCS. For detailed information, please refer to Appendix 2.

Survey Development

The list of attributes and attribute levels based on the literature and interview findings included efficacy endpoints (pCR, DFS, OS), impact of side effects on QoL, and BCS, as opposed to a mastectomy. These attributes and attribute levels were finalized with PAG and HCP input, to be included in the DCE (Table 1).

Table 1

Attributes and Associated Levels

Attribute	Level
Pathological complete response (% of patients for whom there are no more invasive cancer cells found in the resected breast and lymph nodes in the region after treatment)	25%
	50%
	75%
	100%
Disease-free survival (% of patients for whom the disease has not come back 5 y after treatment)	25%
	50%
	75%
	95%
	5-y disease-free data unknown
Overall survival (% of patients who will be alive 5 y after treatment)	50%
	65%
	80%
	95%
	5-y survival data unknown
Impact of side effects on QoL (impact of treatment side effects on everyday activities, e.g., looking after yourself, including self-care, and overall QoL)	Mild side effects that do not limit everyday activities
	Moderate side effects with occasional limits on everyday activities
	Severe side effects but are not life-threatening with some limits on everyday activities
Breast-conserving surgery (surgical treatment that removes cancer from breast while leaving as much healthy breast tissue as possible; also known as lumpectomy)	Yes
	No

QoL, quality of life.

Combinations of attributes and attribute levels for the profiles in the DCE tasks were generated using an experimental design. A statistically D-efficient main effects design was estimated using Sawtooth Lighthouse Studio software (version 9.14.2), resulting in a total of 750 experimental DCE tasks. These tasks were partitioned into 50 blocks, each comprising 15 choice tasks. Patients were randomly assigned to 1 of the 50 blocks and completed the 15 choice tasks.

For each task, patients were asked which treatment option they would choose in the neoadjuvant HER2+ eBC setting. Each task presented patients with a choice between 3 hypothetical alternatives: treatment profile A, treatment profile B, and an “opt-out” option. The treatment profile options in the choice tasks were defined by the 5 attributes and varying attribute levels, with associated definitions to ensure language was clear to patients. Infeasible combinations of attributes and attribute levels were excluded when generating the DCE design (please see Table S4 in Appendix 3 for further information). Choice tasks were tested with PAGs and HCPs prior to survey launch. Figure 1 provides an example choice task.

Figure 1

Illustrative discrete choice experiment task.

The 20-min survey also included sociodemographic and clinical characteristic questions and questions regarding the ease of task completion.

The survey was pilot tested with 50 patients to evaluate whether patients understood the choice tasks. The survey and design were developed following International Society for Pharmacoeconomics and Outcomes good research practices for DCE studies.¹⁹

Participants

Patients were recruited between October and December 2022. Patients were included if they were diagnosed with either hormone receptor–positive (HR+)/HER2+ or hormone receptor–negative (HR−)/HER2+ BC in the early stage (stage 0, I, IIA, IIB, IIIA), had previously received or planned to undergo neoadjuvant treatment, or had received neoadjuvant treatment in the past 5 y if they had received any prior treatment. Other criteria included being at least 18 y old, access to an internet browser or application, and ability to interpret the survey.

The sample size required for assessing the main effects for preference weights depends on the number of choice tasks (t), number of alternatives (a), and number of analysis cells (c) equal to the largest number of levels for any of the attributes: N > 500c/(t×a).^22,23 This allowed for 15 choice tasks, 5 levels, with a suggested sample of 85 patients per country.

Statistical Analyses

The DCE data were pooled and analyzed across countries using multinomial logit regressions (MNL) and random parameters logit (RPL) models.

Prior to the main analysis, addressing multicollinearity is crucial in regressions, as high correlations can lead to models with unstable estimates and erroneous significant levels.²⁴ Multicollinearity issues are likely to arise when correlations between pairs of independent variables exceed 0.7.²⁴ Pairwise correlations among attributes and attribute levels were analyzed to assess the evidence of multicollinearity prior to the remaining analyses.

Choices for attributes and attribute levels were included in MNL models to estimate preference weight coefficients, using hierarchical Bayes estimation. Preference weights (log-odd coefficients) measure the relative effect of an attribute level on the utility or preference for a hypothetical eBC treatment. The patient’s utility (U) (the dependent variable) associated with an alternative was described by a nonexplainable random component (ε) and a systematic utility component. The following utility function for treatment alternatives was specified, with the “no treatment” alternative as the reference alternative:

\begin{matrix} U_{i t j} = α_{t r e a t m e n t . a} + α_{t r e a t m e n t . b} + β_{p c r 2} X_{1 i t j} + β_{p c r 3} X_{2 i t j} \\ + β_{p c r 4} X_{3 i t j} + β_{d f s 1} X_{4 i t j} + β_{d f s 2} X_{5 i t j} + β_{d f s 3} X_{6 i t j} \\ + β_{d f s 4} X_{7 i t j} + β_{o s 1} X_{8 i t j} + β_{o s 2} X_{9 i t j} + β_{o s 3} X_{10 i t j} \\ + β_{o s 4} X_{11 i t j} + β_{s i d e e f f e c t s 1} X_{12 i t j} + β_{s i d e e f f e c t s 2} X_{13 i t j} \\ + β_{b c s 1} X_{14 i t j} + ε_{i t j} \end{matrix}

where a patient ( $i = 1, \dots, n)$ chose an alternative $(j = 1, 2, 3)$ in each choice set $(t = 1, \dots, 15)$ with the following attributes: PCR $(β_{2} X_{1 i t j} \dots β_{4} X_{3 i t j}),$ DFS $(β_{1} X_{4 i t j} \dots β_{4} X_{7 i t j})$ , OS $(β_{1} X_{8 i t j} \dots β_{4} X_{11 i t j})$ , impact of side effects on QoL $(β_{1} X_{12 i t j}, β_{2} X_{13 i t j})$ , BCS $(β_{1} X_{14 i t j})$ .

A dummy-coded specification of attribute levels (i.e., effect of change from the reference level to another level) was used. Compared with effects coding, dummy coding is easier to interpret and avoids issues of misspecification and misinterpretation, although it can create correlations between estimated coefficients of levels for a categorical attribute and the model constant.^25–28

A linearly coded version of the MNL model was also estimated to quantify patient preferences for a 1-unit change (1% increase) in the pCR attribute. The incremental changes in pCR attribute levels (25%, 50%, 75%, 100%) were then derived from the 1% increase estimate. The linearly coded version of the model assumes attribute levels for pCR are presented as continuous numerical values rather than categorical indicators. This tested the assumption that a unit change in utility has a constant effect on preferences and does not depend on the absolute value of the attribute level for pCR. Other attributes were dummy coded. An RPL model, with normal distributions for attribute levels and fixed distributions for the alternative specific constant and 3,000 draws from pseudo random estimation, was used to account for differences in preferences.²⁹ The comparison of statistical performance across models was based on goodness-of-fit statistics, including Bayesian information criterion (BIC), adjusted McFadden pseudo R² (APR), and log-likelihood. The variation in preferences was evaluated between subgroups associated with observable characteristics.

Next, the conditional relative attribute importance (RAI) of an attribute over the range of levels was computed. Preference weight coefficients for each attribute level were normalized, allowing for comparisons and assessing the conditional RAI from most important to least important on a 0 to 100 scale. The reference level was set to the least important value. Differences in preference weights between the most preferred attribute level and the least preferred level of the same attribute provided an estimate of the relative importance of that attribute over the range of levels.

RA I_{k} = 100 \times \frac{{\hat{β}}_{k}^{\max} - {\hat{β}}_{k}^{\min}}{\sum_{k \in [1; K]} ({\hat{β}}_{k}^{\max} - {\hat{β}}_{k}^{\min})}

where $({\hat{β}}_{k}^{\max} - {\hat{β}}_{k}^{\min})$ corresponds to the largest change in utilities obtained with moving from the worst to best level of the $k^{th}$ attribute. If the worst level also happened to be the reference category for the dummy coding of the effects, then ${\hat{β}}_{k}^{\min} = 0$ . The Delta method was used to obtain standard errors and build 95% confidence intervals.²⁹

Patients were asked to rank attributes by level of importance, where 1 = most important and 5 = least important. Patients were also asked how easy or difficult it was to complete the DCE and whether they understood the DCE and relevant attributes.

Statistical analyses were conducted using SAS® 9.2. Separate analyses were also conducted in R 4.2.0 for quality control.

Results

Patient Characteristics

The sample (Table 2) included 334 patients (Germany: 83; France: 85; Italy: 83; Spain: 83) who met the inclusion criteria, provided informed consent, and whose surveys were considered complete. The median time for survey completion was 11 min. The mean age was 40.1 y (range, 18.0–74.0 y). The sample consisted predominantly of White/Caucasian individuals (92%) and those who reported having an undergraduate degree (40%). Approximately 38% of patients received a diagnosis between 6 mo and 1 y ago, with a tumor found only in the breast (54%). Most patients also previously received treatment for eBC (84%) and planned to receive further treatment for eBC (89%). For those planning to receive treatment (n = 298), most planned to receive chemotherapy (52%), followed by HER2-targeted therapy (40%). For patients who previously received surgery (n = 145), chemotherapy was the most common treatment prior to surgery (58%), followed by HER2-targeted therapy (43%). Approximately 75% of patients received or were planning to receive treatment (adjuvant treatment) after they received surgery (n = 252), with chemotherapy (42%) and HER2-targeted therapy (36%) as the most common treatments. More than half of patients (56%) reported testing positive for the BRCA gene mutation/other genetic predispositions.

Table 2

Patient Characteristics

Characteristic	Overall Sample (N = 334)
Age, y
Mean (SD)	40.1 (10.0)
Min; Max	18.0; 74.0
Sex, n (%)
Male	5 (1.5)
Female	329 (98.5)
Other	0 (0)
Highest education level, n (%)
Secondary education (pre–high school or completed at age 15/16 y)	11 (3.3)
High school	48 (14.4)
Trade/technical vocational training	48 (14.4)
Undergraduate education (bachelor’s degree; DUT for technical vocational training)	135 (40.4)
Postgraduate education (master’s degree, PhD)	92 (27.5)
Ethnic group, n (%)
White (including any White ethnic group)	306 (91.6)
Mixed/multiple ethnic groups	7 (2.1)
Asian (including Pakistani, Indian, Bangladeshi, Chinese or any Asian ethnic group)	16 (4.8)
Black/African/Caribbean (including any Black, African, or Caribbean ethnic group)	3 (0.9)
Middle East (including Arab, Persian, Kurd or any Middle East ethnic group)	2 (0.6)
Marital status, n (%)
Single (never married)	71 (21.3)
Married or civil partnership	237 (71.0)
Married but separated	11 (3.3)
Widow and not remarried	1 (0.3)
Divorced and not remarried	14 (4.2)
Employment status, n (%)
Working (full-time)	245 (73.4)
Working (part-time)	38 (11.4)
Self-employed	9 (2.7)
Not working/unemployed	25 (7.5)
Retired	11 (3.3)
Relatives have previously been diagnosed with breast cancer, n (%)
Yes	209 (62.6)
No	119 (35.6)
Unsure	6 (1.8)
Relative(s) was diagnosed with breast cancer,^a,b n (%)
Mother	124 (37.1)
Father	4 (1.2)
Sibling	27 (8.1)
Aunt	54 (16.2)
Uncle	1 (0.3)
Cousin	10 (3.0)
Other	15 (4.5)
Missing	125 (37.4)
Time since diagnosis, n (%)
<6 mo ago	39 (11.7)
Between 6 mo and 1 y ago	127 (38.0)
Between 1 and 2 y ago	84 (25.1)
Between 2 and 3 y ago	56 (16.8)
Between 3 and 4 y ago	18 (5.4)
Between 4 and 5 y ago	10 (3.0)
Have previously received treatment for early-stage breast cancer, n (%)
Yes	280 (83.8)
No	54 (16.2)
Have previously received surgery for early-stage breast cancer or plan to undergo surgery for early-stage breast cancer, n (%)
I have had surgery in the past for early-stage breast cancer	145 (43.4)
I plan to have surgery for early-stage breast cancer	189 (56.6)
Tested positive for BRCA genes mutation or any other genetic predispositions for cancer, n (%)
Yes	186 (55.7)
No	124 (37.1)
Unsure	24 (7.2)
Plan to receive treatment for early-stage breast cancer, n (%)
Yes	298 (89.2)
No	36 (10.8)
Treatment(s) planning to receive for early-stage breast cancer,^a,c n (%)
Chemotherapy	172 (51.5)
HER2-targeted therapy	132 (39.5)
Hormone therapy	72 (21.6)
Radiotherapy	80 (24.0)
Surgery	75 (22.5)
Other	2 (0.6)
Missing	36 (10.8)
Treatment(s) received prior to your surgery,^a n (%)
Chemotherapy	195 (58.4)
HER2-targeted therapy	143 (42.8)
Hormone therapy	80 (24.0)
Radiotherapy	77 (23.1)
Have received or are receiving any treatment after surgery (e.g., adjuvant treatment) for early-stage breast cancer, n (%)
Yes	252 (75.4)
No	82 (24.6)
Current health status, n (%)
Excellent	39 (11.7)
Very good	121 (36.2)
Good	134 (40.1)
Fair	39 (11.7)

BRCA, breast cancer gene; DUT, French National Postsecondary Diploma; Max, maximum; Min, minimum; PhD, doctor of philosophy; SD, standard deviation.

Item response percentages were calculated with the observed n as the denominator.

Not mutually exclusive from each other.

If answered “yes” in “Relatives have previously been diagnosed with breast cancer.”

If answered “yes” in “Plan to receive treatment for early-stage breast cancer.”

Attribute Correlations

Correlations between attributes and attribute levels were analyzed. None of the correlation coefficients between attributes and attribute levels were greater than the 0.7 threshold, indicating no evidence of multicollinearity (Table 3).

Table 3

Attribute Correlation Matrix

	asc.d1	asc.d2	dfs.d1	dfs.d2	dfs.d3	dfs.d4	os.d1	os.d2	os.d3	os.d4	sideff.d1	sideff.d2	bcs.d1	pcr.lin
asc.d1	1.000	0.958	−0.279	−0.307	−0.458	−0.512	−0.471	−0.460	−0.397	−0.306	−0.226	−0.226	−0.175	−0.410
asc.d2		1.000	−0.294	−0.307	−0.464	−0.515	−0.475	−0.457	−0.400	−0.313	−0.231	−0.233	−0.184	−0.411
dfs.d1			1.000	0.488	0.434	0.368	0.001	−0.005	−0.022	0.004	0.036	0.043	−0.007	0.002
dfs.d2				1.000	0.432	0.382	0.016	0.008	0.001	0.039	0.017	0.014	0.001	0.029
dfs.d3					1.000	0.591	0.406	0.407	0.119	0.008	0.013	0.027	0.011	0.021
dfs.d4						1.000	0.507	0.521	0.433	0.024	0.011	0.006	0.009	0.002
os.d1							1.000	0.669	0.581	0.421	−0.002	0.001	−0.003	−0.020
os.d2								1.000	0.578	0.396	0.002	0.000	0.000	−0.057
os.d3									1.000	0.447	−0.014	−0.009	0.008	−0.050
os.d4										1.000	0.024	0.039	0.016	−0.008
sideff.d1											1.000	0.502	0.008	0.019
sideff.d2												1.000	0.007	0.022
bcs.d1													1.000	0.030
pcr.lin														1.000

asc.d1, alternative-specific constant for option A; asc.d2, alternative-specific constant for option B; dfs.d1, disease-free survival for 25% level; dfs.d2, disease-free survival for 50% level; dfs.d3, disease-free survival for 75% level; dfs.d4, disease-free survival for 95% level; os.d1, overall survival for 50% level; os.d2, overall survival for 65% level; os.d3, overall survival for 80% level; os.d4, overall survival for 95% level; sideff.d1, impact of side effects on quality of life for mild side effects level; sideff.d2, impact of side effects on quality of life for moderate side effects level; bcs.d1, breast-conserving surgery for level Yes; pcr.lin, linear-coded PCR.

Treatment Attribute Preference Weights

Figure 2 shows the normalized mean preference weight coefficients for each attribute level. The change in utility associated with a change in each attribute level is represented by vertical differences between preference weight coefficients for attribute levels compared with each reference level. Larger differences between preference weight coefficients mean patients viewed the change having a greater effect on marginal utility.

Figure 2

Attribute-level preference weights for patients with HER2+ eBC (N = 334).

Preferences were ordered as expected, in which patients preferred better attribute levels to worse levels. Most preference weight coefficients for the attribute levels were statistically significant compared with reference levels (P < 0.05), except for 25% DFS compared with unknown DFS data at 5 y and 65% and 50% OS compared with unknown OS data at 5 y.

An RPL model was also estimated to capture heterogeneity in patient preferences. Preferences in the RPL model were also ordered as expected. Given the performance data in the RPL model were similar to the linear-coded MNL (linear-coded MNL: log-likelihood = −4,249.9; BIC = 8,619.1, APR = 22.53%; RPL: log-likelihood = −4,243.8, BIC = 8,709.1, APR = 22.42%), with the linear-coded MNL exhibiting improved fit, the linear-coded MNL was used for the remainder of the analysis. More details on the RPL analysis can be found in Appendix 2.

Conditional Relative Attribute Importance

Figure 3 shows the conditional RAI of each attribute from least preferred to most preferred on a 0 to 100 scale based on the preference weight coefficients. Findings showed pCR was considered the most important attribute relative to all other attributes (31%, SE = 2.50). This change in pCR was estimated to be statistically significantly higher (P < 0.05) than changes in other attributes. This was followed by increases in DFS (24%, SE = 3.57) and OS (22%, SE = 3.04). These were of similar conditional relative importance, while changes in BCS (14%, SE = 1.62) and the impacts of side effects on QoL (11%, SE = 2.06) were less important to patients.

Figure 3

Mean conditional RAI for treatment of eBC (N = 334).

Rankings of Attributes

Most patients ranked pCR most important (first) when making an eBC treatment decision in the neoadjuvant setting (32%). A large proportion of patients also ranked BCS as least important (fifth) (34%).

Subgroup Analyses

Subgroup analyses were conducted to assess whether there were differences in preferences between patient subgroups. Four subgroups were included for final reporting showing statistically significant differences in preferences (using Z-tests to determine if conditional RAI differed significantly between subgroups from the overall sample). These included HR status, time since diagnosis, cancer stage, and age (Figure 4).

Figure 4

Relative attribute importance (%) by sociodemographic and clinical subgroups (N = 334).

HR+/ HER2+ patients (n = 216) placed the most relative importance on pCR compared with all other attributes (32%), whereas DFS was most important to HR−/HER2+ patients (n = 118).

Patients who were either diagnosed less than 1 y ago (n = 166) or between 1 and 3 y ago (n = 140) placed the most relative importance on pCR relative to all other attributes (<1 y: 34%; 1 to 3 y: 28%). Patients who were diagnosed more than 3 y ago (n = 28) placed the most relative importance on OS (48%). Across all groups, pCR, DFS, and OS were most important for eBC treatment decision making in the neoadjuvant setting.

Patients whose tumor was found only in the breast (n = 180) placed the most relative importance on pCR relative to all other attributes (31%). Patients whose tumor was found in the breast and/or lymph nodes (n = 154) placed the most relative importance on DFS (30%). pCR, DFS, and OS were considered the most important attributes in these groups.

Patients aged 18 to 29 y (n = 53), 50 to 59 y (n = 28), and 60 y and older (n = 11) considered OS most important relative to all other attributes (18 to 29 y: 37%; 50 to 59 y: 33%; ≥60 y: 30%). Patients aged 30 to 39 y (n = 82) placed the most importance on pCR (29%). DFS was considered most important to patients aged 40 to 49 y (n = 160) (29%). pCR, DFS, and OS were considered the most important attributes across the different age groups.

Face and Content Validity of Preferences

The ease of selecting choices in the DCE and comprehension of attributes were analyzed. Overall, 52% of patients considered ease of selecting choices “somewhat easy,” followed by “very easy” (21%). Most also either fully understood (77%) or somewhat understood (22%) the DCE scenarios, with only 1 patient (1%) not understanding the scenarios. The attributes were either fully understood (73%) or somewhat understood (26%) by most patients. Only 3 patients (1%) did not understand the attributes, with pCR, DFS, and BCS selected by 1 patient each. However, when removing these 3 patients from the analysis, there were no significant differences in preferences compared with the remaining sample, so these patients were included in the analysis.

Furthermore, nontrading behavior, in which participants always choose the same alternatives across tasks, was analyzed.³⁰ Only 1.20% (n = 4) of patients always chose treatment profile A, 1 patient (0.3%) always chose treatment profile B, and none of the patients (0%) always chose the opt-out alternative. These 5 patients exhibiting nontrading behavior throughout the DCE were removed initially to see if their choices affected preferences (Appendix 4). Findings showed preferences with the reduced sample (n = 329) were similar to the original sample (N = 334), so the patients were included in the analysis.

Discussion

The aim of this study was to quantify preferences for treatment attributes and endpoints with patients with HER2+ eBC in the neoadjuvant setting in Europe. Attributes were selected based on the published literature and qualitative discussions with PAG members and HCPs. Findings from the TLR and discussions provided an understanding of what patients with HER2+ eBC value when evaluating treatments in the neoadjuvant setting.

Patients with HER2+ eBC in this sample placed the most value on pCR, DFS, and OS. The impact of the side effects on QoL and BCS were less important to patients. The change in pCR from 25% to 100% was most important to patients when making treatment decisions for eBC. DFS and OS were also important to patients, with greater marginal utility placed on increases in DFS and OS at 5 y. Utility decrements between attribute levels were not always linear for every endpoint, pointing toward diminishing returns for patients. There were steep drops in marginal utility between the second and third highest attribute levels for OS and the impact of side effects on QoL. There were fewer differences in marginal utility between the first and second highest attribute levels for OS, in which patients considered 95% OS only slightly better than 80% compared with data unknown. For DFS, there was a steep drop in marginal utility from the first and second highest attribute levels but a minimal difference between the second (50%) and third (75%) attribute levels. For pCR, decrements between the levels were similar, showing linear decreases in preference across each attribute level. Results from rankings mirrored the DCE, with most patients ranking pCR as most important, followed by DFS and OS, and placing less importance on BCS and the impacts of side effects on QoL.

Between subgroups, pCR was most important to patients aged 30 to 39 y, although DFS and OS were also important for treatment decision making. Research shows younger patients (<40 y) are more likely to receive intensive treatments and therefore prioritize tumor removal to achieve pCR.³¹ Younger patients may also place importance on removing the tumor due to fertility preservation. Standard breast cancer treatments can negatively affect reproductive health resulting in delays in childbearing and inability to breastfeed, so this age group may prioritize tumor removal to achieve pCR.³² Tumor location can also have an impact on a patient’s psychological outlook. For patients whose tumor was found only in the breast, pCR was most important, but for those whose tumor was found in the breast and lymph nodes, DFS was most important. Patients with localized breast tumors may have a positive outlook on their prognosis from the tumor being removed, while those with spread tumors may be more focused on remaining disease free for a longer period of time.³³ OS was considered most important for those whose time since diagnosis was greater than 3 y. Since achieving pCR is associated with earlier response to neoadjuvant treatment and those with longer time since diagnosis are more likely to have experienced recurrence, these patients may place more value on longer-term outcomes.³⁴ HR+/ HER2+ patients also placed most importance on pCR, whereas HR−/HER2+ patients considered DFS most important. Familiarity with these endpoints may depend on what treatment patients received or what was discussed with physicians. These findings showed significant differences in preferences between subgroups of patients with HER2+ eBC, with patients considering pCR, DFS, and OS as key attributes for treatment decision making. However, given the large imbalance between the number of patients for each subgroup and the small sample size for some subgroups (n≤20), results should be interpreted with caution.

In addition to OS, pCR and DFS were important to patients, indicating that in some situations, HER2+ eBC patients would be willing to trade off a greater probability of longer OS for the probability of achieving pCR. Although these findings were surprising, there may be a few explanations. Patients may view OS outcomes both as a chance of survival and a risk of nonsurvival (e.g., even a 95% OS at 5 y constitutes 5% of patients at risk for nonsurvival) and prefer the certainty of complete tumor removal over the uncertainty of what happens between diagnosis and survival. Although high pCR rates tend to be a positive indicator of long-term outcomes, achieving pCR does not always guarantee high OS or eliminate all risks, particularly given the aggressive nature of HER2+ eBC. In addition, pCR may be understood better as it relates to patients’ experiences of responding to neoadjuvant treatment and provides evidence of the disease being eliminated from the body. Patients may also see a relationship between the efficacy endpoints and consider pCR as potential “stepping stones” to DF and OS. These findings are consistent with previous literature, noting pCR, DFS, and OS as important endpoints for patient decisions in the neoadjuvant setting.^21,35,36 These findings are also in line with reviews of regulatory bodies.^{8,21,35,37,38} Lastly, these results reflect findings from earlier qualitative work, in which PAG members and HCPs highlighted that in addition to OS, pCR and DFS are important efficacy endpoints. However, EFS was considered less relevant for patient decisions based on the TLR and qualitative discussions. Patients may require additional understanding of EFS and how it differs from DFS. As EFS is listed as a potential surrogate for OS, further discussions are needed with patients and HCPs to disentangle these relationships between pCR, DFS/EFS, and OS in HER2+ eBC in the neoadjuvant setting.

There are several strengths in the current study. One was the use of a DCE to elicit preferences for eBC treatments, following good research practices conducting patient preference studies.^11,39,40 The benefit of a DCE is that it asks patients to assess the relative importance of attributes and make tradeoffs by reflecting on different combinations of attributes and attribute levels for potential treatments.⁴¹ This provides an understanding of key factors that could influence treatment decision making for eBC in the neoadjuvant setting, simulating real-world decisions. This study also used both qualitative insights and quantitative approaches to conceptually and analytically integrate eBC attributes. Another strength was the inclusion of preference heterogeneity and subgroup analyses, with potential clinical implications for individual treatment decision making. Feedback from HCPs and PAG members was incorporated throughout the course of the study, including qualitative interviews, PAG advisory board discussions, and review of study documents. Lastly, the opt-out alternative represents the status quo, where patients choose neither alternative A treatment nor alternative B treatment in the hypothetical scenarios. Including an opt-out alternative avoids forcing patients to choose alternatives they do not prefer and is a representation of real-life choice situations.⁴² The absence of a constant preference for an opt-out suggests patients actively considered the tradeoffs between attributes.

These findings should be interpreted with caution in the context of several limitations. All data were self-reported from this sample, and there is risk of information bias, as patients may have given inaccurate responses. There may also be the possibility of unintentional recall bias with self-reported data, as some patients had previously undergone neoadjuvant treatment up to a 5-y period. Best efforts were made to minimize this using patient-friendly wording and sufficient time for survey completion. Second, because study participation was voluntary, the sample may not reflect the eBC disease population, resulting in selection bias. Information was retained for patients not meeting selection criteria for enrollment. Patient characteristics were also described per the screener, and patients who did not meet the eligibility criteria did not move forward to complete the survey. Another point regarding the representativeness of the sample is that most patients completed a higher education degree, which has been shown to positively correlate with higher health literacy.⁴³ Exploring patient preferences among patients with lower education degrees may yield different results. Furthermore, the sample did not proportionately capture older patients (older than 50 y of age), despite older individuals comprising most eBC. Nevertheless, the study reflects the younger, HER2+ patient population, and results demonstrate implications for younger patients younger than 50 y. This may be a benefit with an underrepresented and understudied group; thus, the findings provide implications for younger individuals with eBC. The sample also consisted predominantly of White/Caucasian patients and in Italy consisted of White/Caucasian patients only. This limits the generalizability of these findings to the wider population. Future work should focus on cross-validating preferences with other ethnic groups and languages. The study design can also impose a limitation to real-world applicability, as patients were asked to choose between 2 hypothetical treatment options plus an opt-out alternative. It is unclear how patients would make these choices in the real world, as they could respond differently in a clinical setting. There may also be some social desirability bias when completing the choice exercises, which could have contributed to these findings. In addition, there may be treatments important to patients with HER2+ eBC that were not included in the survey. Attributes were selected based on the TLR and discussions with HCPs and PAG members, some of whom were previous eBC patients. Attribute levels were also selected using a hybrid approach of matching treatments from the literature and expanding to potential future treatments for HER2+ eBC, so some levels may be more hypothetical in nature. Although attribute levels were tested with PAGs and HCPs prior to survey launch, they may not fully reflect all possible treatment options, which has likely influenced preferences. Future research should explore preferences for HER2+ eBC treatments with an independent sample. Finally, a small proportion of patients reported not understanding certain attributes. To address this, patients were asked about their understanding of attributes. Understanding of the DCE also showed most patients had an overall good comprehension of the tasks and attributes.

There are several implications drawn from this study. To our knowledge, this is the first time preferences have been estimated for patients with HER2+ eBC in the neoadjuvant setting, capturing not only the importance of OS but also of surrogate endpoint measures of OS for treatment decision making. These findings are in line with views of regulatory bodies, which recently accepted pCR and DFS as clinically validated endpoints, particularly in high-risk eBC.^8,44 Physicians also reported pCR as one of the most important endpoints to assess the success of neoadjuvant therapy.⁴⁵ Most notably, these findings reflect what matters most to patients, which can be used to align care for HER2+ eBC patients in the neoadjuvant setting. Given the early stage of the disease, the use of surrogate endpoints and other measures demonstrating patient benefit in the short term may be beneficial for timely assessment of eBC treatments and faster patient access. Payers should consider these endpoints when making decisions about reimbursement for eBC treatments. The findings do not, however, imply conclusions about the extent non-OS endpoints can/should be considered as a surrogate marker for OS. Additional work is critical to better understand the relationships between these endpoints, capture any other endpoints not tested (e.g., real-world clinical data), and assess their importance to patients.

The current sample of patients with HER2+ eBC placed importance on pCR, DFS, and OS when considering treatments in the neoadjuvant setting. These findings provide insights into the importance of endpoints and the tradeoffs patients would be willing to make when choosing neoadjuvant eBC treatments. This helps payers understand the value of non-OS endpoints to patients, especially for early disease.

Supplemental Material

sj-docx-1-mpp-10.1177_23814683251372622 – Supplemental material for Patient Preferences in Neoadjuvant Therapy for HER2+ Early-Stage Breast Cancer

Supplemental material, sj-docx-1-mpp-10.1177_23814683251372622 for Patient Preferences in Neoadjuvant Therapy for HER2+ Early-Stage Breast Cancer by Laurie Batchelder, Laure Guéroult-Accolas, Eirini Anastasaki, Kyle Dunton, Diana Lüftner, Corinna Oswald, James Ryan, Doris C. Schmitt, Veronika Steinerova, Della Varghese and Sukhvinder Johal in MDM Policy & Practice

Footnotes

Acknowledgements

The authors would like to thank all the patients who participated in this study; Giampaolo Bianchini of the Department of Medical Oncology in Ospedale San Raffaele, Italy, for providing thoughtful clinical insights on the study documents; David Alsadius of IQVIA, Sweden, for providing clinical implications throughout the study; Dian Zhang of IQVIA, Virginia, for statistical support; Ana Maria Rodriguez of IQVIA, Spain, for scientific oversight of the study duration and the study documents’ scientific rigor following good research practices for preference research; Ana Sofia Oliveira Gonçalves of IQVIA, Portugal, for providing medical writing support; and Medefield, IQVIA’s third-party recruitment vendor, for recruiting patients and health care providers.

The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: JR, DV, and SJ are employees of AstraZeneca, who co-funded this study. KD is an employee of Daiichi Sankyo Europe GmbH, who co-funded this study. LB, EA, CO, and VS are employees of IQVIA, who were contracted to perform this research on behalf of AstraZeneca and Daiichi Sankyo Europe GmbH. The abstract of this article was presented at ISPOR Europe 2023 as a poster presentation with initial results. The poster’s abstract was published in “Poster Abstracts” in Value in Health, volume 26, issue 12, supplement, S452–S453, December 2023, DOI: . The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was funded by AstraZeneca UK Limited, in collaboration with Daiichi Sankyo Europe GmbH. The research was conducted by IQVIA on behalf of AstraZeneca and Daiichi Sankyo.

Ethics Approval and Informed Consent

This study was reviewed and approved through an exemption regulatory opinion by WCG IRB 1019 39th Ave, SE Suite 120 Puyallup, WA 983374 (dated September 27, 2022). The study protocol, informed consent, screener, survey, and recruitment materials were reviewed and approved prior to implementation. All patients provided their informed consent to participate in this study.

ORCID iD

Laurie Batchelder

Data Availability

As stated in the approved study protocol, only members of the research team (study authors) have access to the study data. The full anonymized data set was shared between all team members. Direct access will be granted for monitoring and/or audit of the study to ensure compliance with regulations.

References

Arnold

Morgan

Rumgay

, et al. Current and future burden of breast cancer: global statistics for 2020 and 2040. Breast. 2022;66:15–23. DOI:10.1016/j.breast.2022.08.010

National Cancer Institute. Definition of early-stage breast cancer. In: NCI Dictionary of Cancer Terms. 2011. Available from: https://www.cancer.gov/publications/dictionaries/cancer-terms/def/early-stage-breast-cancer. [Accessed 23 March, 2023].

Wuerstlein

Harbeck

. Neoadjuvant therapy for HER2-positive breast cancer. Rev Recent Clin Trials. 2017;12(2):81–92. DOI:10.2174/1574887112666170202165049

Harbeck

. Neoadjuvant and adjuvant treatment of patients with HER2-positive early breast cancer. Breast. 2022;62(suppl 1):S12–6. DOI:10.1016/j.breast.2022.01.006

Gion

Trapani

Cortés

, et al. Systemic therapy for HER2-positive metastatic breast cancer: moving into a new era. Am Soc Clin Oncol Educ Book. 2022;42:82–92. DOI:10.1200/EDBK_351222

Driscoll

Rixe

. Overall survival: still the gold standard: why overall survival remains the definitive end point in cancer clinical trials. Cancer J. 2009;15(5):401. DOI:10.1097/PPO.0b013e3181bdc2e0

Lux

Ciani

Dunlop

Ferris

Friedlander

. The impasse on overall survival in oncology reimbursement decision-making: how can we resolve this? CMAR. 2021;13:8457–71. DOI:10.2147/CMAR.S328058

Gion

Pérez-García

Llombart-Cussac

Sampayo-Cordero

Cortés

Malfettone

. Surrogate endpoints for early-stage breast cancer: a review of the state of the art, controversies, and future prospects. Ther Adv Med Oncol. 2021;13:17588359211059588. DOI:10.1177/17588359211059587

Duranti

Fabi

Filetti

, et al. Breast cancer drug approvals issued by EMA: a review of clinical trials. Cancers (Basel). 2021;13(20):5198. DOI:10.3390/cancers13205198

10.

US Food and Drug Administration. Patient preference information: voluntary submission, review in premarket approval applications, humanitarian device exemption applications, and de novo requests, and inclusion in decision summaries and device labeling. 2016. Available from: https://www.fda.gov/media/92593/download. [Accessed 16 August, 2023].

11.

IMI PREFER. CHMP & EUnetHTA parallel scientific advice: qualification of a framework and “points to consider” for method selection along with five methods for performing patient preference studies to inform regulatory and HTAbody medical product decision-making. 2021. Available from: https://www.ema.europa.eu/en/documents/regulatoryprocedural-guideline/chmp-eunethta-parallel-scientific-advice-qualification-framework-points-considermethod-selection_en.pdf.

12.

Krist

Tong

Aycock

Longo

. Engaging patients in decision-making and behavior change to promote prevention. Stud Health Technol Inform. 2017;240:284–302.

13.

Delevry

. Effect of treatment preference in randomized controlled trials: systematic review of the literature and meta-analysis. Patient. 2019;12(6):593–609. DOI:10.1007/s40271-019-00379-6

14.

Gyawali

Hey

Kesselheim

. Evaluating the evidence behind the surrogate measures included in the FDA’s table of surrogate endpoints as supporting approval of cancer drugs. EClinicalMedicine. 2020;21:100332. DOI:10.1016/j.eclinm.2020.100332

15.

Kong

Moran

Zhang

Haffty

Yang

. Meta-analysis confirms achieving pathological complete response after neoadjuvant chemotherapy predicts favourable prognosis for breast cancer patients. Eur J Cancer. 2011;47(14):2084–90. DOI:10.1016/j.ejca.2011.06.014

16.

Spring

Fell

Arfe

, et al. Pathological complete response after neoadjuvant chemotherapy and impact on breast cancer recurrence and survival: a comprehensive meta-analysis. Clin Cancer Res. 2020;26(12):2838–48. DOI:10.1158/1078-0432.CCR-19-3492

17.

Binotto

Reinert

Werutsky

Zaffaroni

Schwartsmann

. Health-related quality of life before and during chemotherapy in patients with early-stage breast cancer. Ecancermedicalscience. 2020;14:1007. DOI:10.3332/ecancer.2020.1007

18.

Mokhtari-Hessari

Montazeri

. Health-related quality of life in breast cancer patients: review of reviews from 2008 to 2018. Health Qual Life Outcomes. 2020;18(1):338. DOI:10.1186/s12955-020-01591-x

19.

Reed Johnson

Lancsar

Marshall

, et al. Constructing experimental designs for discrete-choice experiments: report of the ISPOR conjoint analysis experimental design good research practices task force. Value Health. 2013;16(1):3–13. DOI:10.1016/j.jval.2012.08.2223

20.

Bridges

JFP

Hauber

Marshall

, et al. Conjoint analysis applications in health—a checklist: a report of the ISPOR good research practices for conjoint analysis task force. Value Health. 2011;14(4):403–13. DOI:10.1016/j.jval.2010.11.013

21.

Thill

Pisa

Isbary

. Targets for neoadjuvant therapy—the preferences of patients with early breast cancer. Geburtshilfe Frauenheilkd. 2016;76(5):551–6. DOI:10.1055/s-0042-101025

22.

Johnson

Orme

. Getting the most from CBC (2003). 2003. Available from: https://sawtoothsoftware.com/resources/technical-papers/getting-the-most-from-cbc. [Accessed 21 April, 2023].

23.

de Bekker-Grob

Donkers

Jonker

Stolk

. Sample size requirements for discrete-choice experiments in healthcare: a practical guide. Patient. 2015;8(5):373–84. DOI:10.1007/s40271-015-0118-z

24.

Mark

Swait

Using stated preference and revealed preference data fusion modelling in health care. In: Ryan

Gerard

Amaya-Amaya

, eds. Using Discrete Choice Experiments to Value Health and Health Care. Dordrecht (the Netherlands): Springer; 2008. p 217–34.

25.

Balogh

Békési

Gorton

Popp

Lengyel

. Consumer willingness to pay for traditional food products. Food Policy. 2016;61:176–84. DOI:10.1016/j.foodpol.2016.03.005

26.

Bech

Gyrd-Hansen

. Effects coding in discrete choice experiments. Health Econ. 2005;14(10):1079–83. DOI:10.1002/hec.984

27.

Hoyos

. The state of the art of environmental valuation with discrete choice experiments. Ecol Econ. 2010;69(8):1595–603. DOI:10.1016/j.ecolecon.2010.04.011

28.

Sun

Penn

Qing

. Dummy and effects coding variables in discrete choice analysis. Am J Agric Econ. 2022;104(5):1770–88. DOI:10.1111/ajae.12311

29.

Hensher

Greene

. The mixed logit model: the state of practice. Transportation. 2003;30:133–76.

30.

Hess

Rose

Polak

. Non-trading, lexicographic and inconsistent behaviour in stated choice data. Transp Res D Transp Environ. 2010;15(7):405–17. DOI:10.1016/j.trd.2010.04.008

31.

Fredholm

Eaker

Frisell

Holmberg

Fredriksson

Lindman

. Breast cancer in young women: poor survival despite intensive treatment. PLoS One. 2009;4(11):e7695. DOI:10.1371/journal.pone.0007695

32.

Shah

Scott

Kandagatla

, et al. Young women with breast cancer: fertility preservation options and management of pregnancy-associated breast cancer. Ann Surg Oncol. 2019;26(5):1214–24. DOI:10.1245/s10434-019-07156-7

33.

Pan

Wang

Qian

, et al. Comparison of survival outcomes among patients with breast cancer with distant vs ipsilateral supraclavicular lymph node metastases. JAMA Netw Open. 2021;4(3):e211809. DOI:10.1001/jamanetworkopen.2021.1809

34.

Hong

Huang

, et al. Early response and pathological complete remission in Breast Cancer with different molecular subtypes: a retrospective single center analysis. J Cancer. 2020;11(23):6916–24. DOI:10.7150/jca.46805

35.

Lai

Antill

Jung

, et al. 16P Patient and healthcare practitioner preferences in early-stage triple-negative breast cancer treatment: a discrete choice experiment. Ann Oncol. 2023;34:S1472–3. DOI:10.1016/j.annonc.2023.10.148

36.

Beusterien

Maculaitis

Hallissey

Gaschler

Smith

Law

. Patient, oncologist, and payer preferences for adjuvant endocrine therapy and CDK4/6 inhibitor regimens in early-stage breast cancer: a discrete choice experiment. Patient Prefer Adherence. 2021;15:611. DOI:10.2147/PPA.S298670

37.

Center for Drug Evaluation and Research. Pathological complete response in neoadjuvant treatment of high-risk early-stage breast cancer: use as an endpoint to support accelerated approval. US Food and Drug Administration. 2020. Available from: https://www.fda.gov/regulatory-information/search-fda-guidance-documents/pathological-complete-response-neoadjuvant-treatment-high-risk-early-stage-breast-cancer-use. [Accessed 27 March, 2023].

38.

European Medicines Agency. Overview of comments received on “Guideline on the role of the pathological complete response (pCR) as an endpoint in neoadjuvant breast cancer studies” (EMA/CHMP/151853/2014). 2015. Available from: https://www.ema.europa.eu/en/documents/comments/overview-comments-received-guideline-role-pathological-complete-response-pcr-endpoint-neoadjuvant-breast-cancer-studies_en.pdf

39.

Bridges

JFP

de Bekker-Grob

Hauber

, et al. A roadmap for increasing the usefulness and impact of patient-preference studies in decision making in health: a good practices report of an ISPOR task force. Value Health. 2023;26(2):153–62. DOI:10.1016/j.jval.2022.12.004

40.

Hauber

González

Groothuis-Oudshoorn

CGM

, et al. Statistical methods for the analysis of discrete choice experiments: a report of the ISPOR conjoint analysis good research practices task force. Value Health. 2016;19(4):300–15. DOI:10.1016/j.jval.2016.04.004

41.

Marshall

. What are patient preferences, how do you measure patient preferences and how can I apply them in clinical practice for patients? Osteoarthr Cartil. 2019;27:S15–6. DOI:10.1016/j.joca.2019.02.009

42.

Campbell

Erdem

. Including opt-out options in discrete choice experiments: issues to consider. Patient. 2019;12(1):1–14. DOI:10.1007/s40271-018-0324-6

43.

Chang

Lin

. Factors influencing implementation of shared medical decision making in patients with cancer. Patient Prefer Adherence. 2019;13:1995–2005. DOI:10.2147/PPA.S217561

44.

Gourgou-Bourgade

Cameron

Poortmans

, et al. Guidelines for time-to-event end point definitions in breast cancer trials: results of the DATECAN initiative (Definition for the Assessment of Time-to-event Endpoints in CANcer trials). Ann Oncol. 2015;26(5):873–9. DOI:10.1093/annonc/mdv106

45.

von Minckwitz

Huang

Mano

, et al. Trastuzumab emtansine for residual invasive HER2-positive breast cancer. N Engl J Med. 2019;380(7):617–28. DOI:10.1056/NEJMoa1814017

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.16 MB