Impact of 21‐Gene Breast Cancer Assay on Treatment Decision for Patients with T1–T3, N0–N1, Estrogen Receptor‐Positive/Human Epidermal Growth Receptor 2‐Negative Breast Cancer: Final Results of the Prospective Multicenter ROXANE Study

Results of the ROXANE study, evaluating the effect of the 21‐gene recurrence score assay on adjuvant treatment decisions for patients with breast cancer, are reported.

Abstract

Background.

The ROXANE Italian prospective study evaluated the impact of the 21‐gene Recurrence Score (RS) results on adjuvant treatment decision for patients with early breast cancer.

Materials and Methods.

Nine centers participated. Physicians used the RS test whenever unsure about adjuvant treatment recommendation for patients with estrogen receptor‐positive/human epidermal growth receptor 2‐negative, T1–T3, N0–N1 early breast cancer. Pre‐RS and post‐RS treatment recommendations were collected.

Results.

A total of 251 patients were included. N0 patients (61%) showed higher grade (p < .001) and higher Ki67 (p = .001) and were more frequently progesterone receptor negative (p = .012) as compared with N1 patients. RS results were as follows: <11, n = 63 (25.1%); 11–25, n = 143 (57%); and ≥26, n = 45 (17.9%). Higher RS was found in N0 vs. N1 patients (p = .001) and in cases of G3 (p < .001) and higher Ki67 (p < .001). The rate of change in treatment decision was 30% (n = 75), mostly from chemotherapy (CT) plus hormone therapy (CT + HT) to hormone therapy (HT; 76%, n = 57/75). The proportion of patients recommended to CT + HT was significantly reduced from pre‐RS to post‐RS (52% to 36%, p < .0001). CT use reduction was more evident for N1 patients (55% to 27%) than for N0 patients (50% to 42%) and was observed only in cases of RS ≤17.

Conclusion.

Physicians predominantly used the 21‐gene assay in N0 patients with a more aggressive biology or in N1 patients showing more indolent biology. In this selected patient population, the use of RS testing led to a 30% rate of change in treatment decision. In the N1 patient subgroup, the use of RS testing contributed to reduce CT use by more than half.

Implications for Practice.

This study shows that, even in a context in which physicians recommend a high proportion of patients to endocrine treatment alone before knowing the results of the Recurrence Score (RS) assay, the use of the RS test, whenever uncertainty regarding adjuvant treatment recommendation is present, significantly contributes in further reducing the use of chemotherapy, especially for N1 patients.

Introduction

An accurate assessment of risks and benefits is crucial in the adjuvant treatment decision‐making process for patients with early hormone receptor (HR)‐positive/human epidermal growth receptor 2 (HER2)‐negative breast cancer in order to avoid over‐ and undertreatment. Indeed, the relative reduction in the risk of relapse achieved by adding adjuvant chemotherapy (CT) to hormone therapy (HT) translates into different degrees of absolute benefit depending on the baseline risk. Multigene prognostic tests have been developed in order to improve the accuracy of a patient's individual risk estimation and are endorsed by the majority of international guidelines, especially for N0 patients [1], [2], [3], [4], [5]. However, in some countries, including Italy, these tests are not currently reimbursed by the National Health Care System. Under these conditions, it is of relevance to acquire regional data that demonstrate the impact these tests would have in real‐world clinical practice. Because treatment patterns and health system regulations vary substantially across different countries, these studies should be conducted on a local basis.

The 21‐gene Oncotype DX Breast Recurrence Score assay (Genomic Health, Inc., Redwood City, CA) is one of the most studied multigene tests. The test provides a Recurrence Score (RS) result that correlates as a continuous variable with the risk of distant relapse. In the first studies, the following risk categories were defined: RS <18 (low risk), RS 18–30 (intermediate risk), and RS >30 (high risk) [6], [7], [8]. In the large prospective TAILORx trial, more than 10,000 patients with HR‐positive, HER2‐negative, N0, T1–T3 breast cancer were assigned to adjuvant therapy based on RS test results, as follows: RS <11, assigned to HT; RS 11–25, randomized 1:1 to HT or CT + HT; RS >25, assigned to CT + HT. The 1,626 patients with an RS <11 had a very low risk of distant recurrence at 5 and 9 years (0.7% and 3.2%, respectively) with HT alone. Patients with a mid‐range RS (11–25, n = 6,711) showed similar invasive disease‐free survival when treated with HT alone or CT + HT (hazard ratio 1.08, 95% confidence interval 0.94–10.24, p = .26) [9], [10]. This study provides level of evidence 1A for the clinical utility of the RS test in this setting.

A number of decision‐impact studies have assessed the rate of change in adjuvant treatment decision associated with the use of the 21‐gene test in European countries. These studies, mostly conducted prior to the availability of the full TAILORx results, generally showed a treatment decision change of around 30% [11], [12], [13], [14].

We previously reported the results of the first decision impact Italian study (Breast‐DX Italy). In this study, all consecutive patients with estrogen receptor (ER)‐positive, HER2‐negative, N0–N1, T1–T3 early breast cancer who met the protocol‐defined criteria of “intermediate risk” (based on classical clinicopathological factors) were offered the RS test. Among the 250 enrolled patients, we reported a rate of treatment decision change of only 16% (mostly from CT + HT to HT) [15].

Building on this experience, a subsequent study was initiated. The rationale was to assess the impact of the RS test on adjuvant treatment decisions in a scenario in which the test was made available to physicians whenever they were unsure about adjuvant treatment recommendations. This design was conceived to capture real‐world data regarding the impact of RS test use in clinical practice.

Materials and Methods

Study Design

ROXANE is a multicenter, prospective decision‐impact study conducted in nine oncologic centers of the Veneto region. The protocol was approved by the ethical committees of all the centers. All patients provided written informed consent.

The aims of this project were to evaluate the change in treatment recommendation pre‐RS to post‐RS and to describe the characteristic of the patients for whom the RS test was ordered.

Physicians from the participating centers had the opportunity to order the RS test whenever they were unsure regarding treatment recommendation for patients with ER‐positive, HER2‐negative early breast cancer (stage T1–T3 and N0–N1). For each patient undergoing the RS test, the following data were collected: pre‐RS physician's recommendation of adjuvant treatment, RS results, post‐RS physician's recommendation, and the type of adjuvant treatment finally received by the patient.

A final questionnaire (available as supplemental online Table 1) was administered to physicians in order to measure confidence in their post‐RS treatment recommendation and their final perception of test utility.

Pathology Evaluation

All routine pathology evaluations including ER, progesterone receptor (PgR), HER2, grade, and Ki67 were assessed locally. ER and PgR were considered positive in cases of positive immunohistochemical staining in at least 10% of tumor cells. HER2 was assessed locally according to 2013 American Society of Clinical Oncology/College of American Pathologists guidelines [16]. For Ki67, although the use of the same antibody across sites was not requested, most of the cases were evaluated with the MIB1 Dako antibody.

Sample Size

The study was initially planned to be open for enrollment for 1 year. It was expected, during this time period, to enroll up to 400 patients across multiple centers. This estimate was based on the previous experience in the Breast‐DX study.

We aimed to observe a rate of change in treatment decision of at least 25%. A sample size of 400 patients would have allowed determination of a 25% change rate with a 95% confidence interval of 21.2%–28.8% (a confidence interval width of 7.6%).

The rate of accrual was lower than initially estimated; therefore, the final sample size after 1 year of enrollment (from January 2017 to February 2018) was 251 patients. This sample size would allow determination of a 25% change rate with an exact 95% confidence interval of 19.8%–30.8% (width of 11%).

Statistical Analysis

Statistical analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC). Associations between variables were studied with the chi‐squared test or the Kruskal‐Wallis test. McNemar's test was used to investigate whether the proportion of patients recommended to CT + HT changed from pre‐RS to post‐RS. All hypothesis tests were conducted at a two‐sided alpha level of 0.05.

Results

Patients’ Characteristics

From January 2017 to February 2018, physicians requested the RS test for 251 patients, n = 152 (60.6%) N0 and n = 99 (39.4%) N1. The main clinicopathological characteristics are summarized in Table 1. We applied the modified Adjuvant! Online clinical risk classification [17] to our patients’ cohort in order to allow comparison with the population of patients enrolled in the MINDACT and TAILORx trials: 70.9% of ROXANE patients were classified as clinical high risk. Comparison of clinicopathological characteristics between N0 and N1 patients is shown in Table 1. Patients with N0 disease were more likely to be classified as modified Adjuvant! Online low risk as compared with N1 patients (44.1% vs. 6.1%, p < .001), partly because nodal status is one of the factors considered in this classifier to assign patients to risk categories. Moreover, there were substantial differences in biologic tumor characteristics according to nodal status, with N0 patients presenting more frequently with aggressive features such as grade 3 (p < .0001), high Ki67 (p = .0006), and negative PgR (p = .0168) as compared with N1 patients.

Table 1. Patients’ clinicopathologic characteristics: overall population and according to nodal status.

Abbreviations: N0, node negative; N1, node positive (1–3 positive nodes); PgR, progesterone receptor; T, tumor size.

Supplemental online Table 2 shows the comparison of clinicopathological characteristics according to menopausal status: Lower Ki67 and positive PgR status were associated with postmenopausal status, and in contrast, premenopausal patients more frequently showed negative PgR and higher Ki67. These data indicate that physicians identified for the RS assay use predominantly N0 and/or postmenopausal patients with more aggressive tumor biology and N1 and/or premenopausal patients as more clinically challenging cases deserving the use of the RS test.

RS Results

RS results according to clinicopathological characteristics are reported in Table 2. Overall, 63 (25.1%) patients had an RS <11, 143 (57.0%) had an RS 11–25, and 45 (17.9%) showed an RS ≥26. A high RS result (RS ≥26) was significantly associated with more aggressive biologic features, such as grade 3 (p < .001), higher Ki67 (p < .001), and negative PgR (p < .001). There was no association of RS results with histologic subtype, tumor size, and modified Adjuvant! Online clinical risk classification. Consistent with the difference in biological characteristics (Table 1), N0 patients more frequently showed a high RS ≥26 as compared with N1 patients (p = .001). Similarly, a higher RS result was observed more frequently in postmenopausal versus premenopausal patients (p = .037).

Table 2. Clinicopathological features according to RS category.

Abbreviations: N0, node negative; N1, node positive (1–3 positive nodes); PgR, progesterone receptor; RS, Recurrence Score; T, tumor size.

Change in Treatment Recommendation

As shown in Figure 1A, 52% of the whole ROXANE population had a pre‐RS treatment recommendation to CT + HT (n = 130); of these patients, 44% had a post‐RS recommendation to HT alone. Of the 121 patients initially recommended to HT alone, 18 (15%) had a post‐RS recommendation to CT + HT. Overall, change in treatment recommendation from pre‐RS to post‐RS occurred in 75 (30%) patients, most frequently from CT + HT to HT alone (n = 57/75, 76%).

Figure 1.

Change in treatment recommendation (pre‐RS➔post‐RS) by pre‐RS recommendation: in the whole population (A), in N0 patients (B), and in N1 patients (C).

Abbreviations: CT, chemotherapy; HT, hormone therapy; RS, Recurrence Score.

Results in N0 and N1 patients are shown in Figure 1B and 1C, respectively. Half of N0 patients (76/152) had a pre‐RS recommendation to CT + HT. Of these patients, 36% had a post‐RS recommendation to HT alone. In addition, 20% of the N0 patients initially recommended to HT alone had a post‐RS recommendation to the addition of CT. In the N0 group, the change in treatment recommendation occurred in 42 patients (28%), and for 64% of them, the change was from CT + HT to HT alone.

In the N1 group, physicians recommended CT + HT pre‐RS to 55% of patients. Of them, 56% had a post‐RS indication to HT alone. The proportion of patients initially recommended to HT alone for whom the recommendation changed to CT + HT was low (7%). Overall, the rate of change in treatment decision was 33% (n = 33) for N1 patients, in the direction of sparing CT (91%, n = 30/33) in the vast majority.

Reduction in CT Recommendation

In the entire ROXANE cohort, a 16% net reduction in the proportion of patients with a recommendation to CT was observed from pre‐RS (52%) to post‐RS (36%), which was statistically significant (p < .001). The reduction in recommendation to CT was observed only in the groups of patients with RS <11 or RS 11–17, as shown in Figure 2A. In the group of patients with RS 18–25, a small increase in CT recommendation occurred (from 43% to 51%). For patients with RS ≥26, the pre‐RS indication to CT was very high (82%) and further increased post‐RS (98%).

Figure 2.

Proportion of patients recommended to HT (blue) or CT + HT (purple) pre‐RS and post‐RS according to RS category: in the whole population (A), in N0 patients (B), and in N1 patients (C).

Abbreviations: CT, chemotherapy; HT, hormone therapy; RS, Recurrence Score.

Similar trends in chemotherapy recommendation changes from pre‐RS to post‐RS according to RS result category were observed when data were analyzed for N0 and N1 patients separately (Fig. 2B, 2C). However, the net reduction in chemotherapy recommendation from pre‐RS to post‐RS was higher for N1 patients (28%, from 55% pre‐RS to 27% post‐RS, p < .001) as compared with N0 patients (8%, from 50% pre‐RS to 42% post‐RS, p = .064). The proportion of patients with recommendation to CT pre‐RS and post‐RS by RS category in subgroups defined by age (≤50 years/>50 years) and by combined age and nodal status are reported in supplemental online Table 3. For all subgroups, the reduction in CT use was evident for RS categories ≤17 and not for higher RS categories.

Type of Recommended CT and HT

Of the 91 patients with a post‐RS recommendation to CT, the type of proposed CT was as follows: taxane‐based (n = 55, 60.4%), anthracycline‐based (n = 14, 15.4%) anthracycline and taxane‐based (n = 14, 15.4%), or other (n = 8, 8.8%). Supplemental online Table 4 shows the changes in CT type recommendation from pre‐RS to post‐RS.

The type of HT recommended post‐RS to postmenopausal patients (n = 142) was aromatase inhibitor (96.5%), tamoxifen (2.8%), or sequential tamoxifen and aromatase inhibitor (0.7%). The type of HT recommended post‐RS to premenopausal patients (n = 105) was aromatase inhibitor + gonadotropin releasing hormone (GnRH) analogue (44.8%), tamoxifen (37.1%), tamoxifen + GnRH analogue (16.2%), or sequential tamoxifen and aromatase inhibitor (1.9%). Supplemental online Table 5 shows the changes in HT type recommendation from pre‐RS to post‐RS.

Treatment Received

The actual treatment received was different from the post‐RS recommended treatment for 15 patients (5.9%): Of these, 12 patients recommended to CT + HT started HT, 2 patients recommended to HT started CT + HT, and 1 patient recommended to HT refused any systemic therapy.

Physicians’ Post‐RS Perception of RS Utility

Overall, physicians declared that they were confident in their treatment recommendation after ordering the RS assay for 74% of the patients; the RS assay provided additional information for 73% of the patients; and the RS results influenced the treatment recommendation for 49% of the patients.

Discussion

The results of the ROXANE study show that the availability of the RS test for use in clinical practice could lead to a change in adjuvant treatment recommendation in up to 30% of patients for whom the clinicians were initially unsure about treatment recommendation based on classical clinical and pathologic factors. This figure is different from the one of our previous Breast‐DX decision impact study (16% change in treatment decision), conducted in the same Italian region [15], and more similar to the results of other European cohorts [11], [12], [13], [14]. There are a number of differences among these studies and factors to be considered when interpreting the results.

We previously discussed the tendency of Italian oncologists to frequently consider HT alone for patients with HR‐positive/HER2‐negative early breast cancer based on the evaluation of classical clinicopathological features, including Ki67 [15]. The population of patients enrolled in the Breast‐DX Italy and in the ROXANE studies showed higher risk features as compared with patients included in other decision impact European studies. As an example, the proportion of N0 patients with grade 3 tumors was higher in the Italian studies (30% in Breast‐DX [15], 49% in ROXANE) as compared with other European studies (13% in the pooled analysis) [11]. Moreover, most European studies focused on N0 patients, and only few included N1 patients, such as Breast‐DX and ROXANE [13], [14], [15]. Nevertheless, in both Italian studies, the rate of pre‐RS recommendation to HT alone was similar or even higher versus other European studies (52% in Breast‐DX, 48% in ROXANE, 40%–55% in other studies) [11], [12], [13], [15]. This prioritization of HT alone for a large proportion of patients may limit the impact of the RS test, which is used, in the majority of cases, in order to identify patients at good prognosis who may be spared CT. This consideration explains why the previous Breast DX study results showed lower rates of treatment changes as compared with European studies but not why the ROXANE results showed a greater impact of the RS test as compared with Breast DX. We hypothesize the main reason for this difference is related to the different design of the two Italian studies. In Breast‐DX, all consecutive patients meeting predefined “intermediate risk” protocol criteria were offered the RS test, irrespective of the physician's perception of RS test utility at baseline. Of the 289 registered patients classified at “intermediate risk” in Breast DX, physicians would not have deemed useful the RS test for 23.5% at baseline [15]. In a clinical context in which the attitude of oncologists toward HT alone is prevalent, a study designed to offer the test to all consecutive patients with predefined features, irrespective of the physician's need, may not fully capture the real impact genomic tests would have if available for clinical routine use. In the ROXANE study, the decision to order the RS test was left at the physician's discretion. The test was available upon request whenever the physician was unsure about treatment recommendation based on classic clinicopathological factors for a patient with ER‐positive/HER2‐negative N0/N1 early breast cancer. This study design mimics a real‐world scenario in which the test is available without any restriction for the population of indication. The characteristics of the patients included in the ROXANE population reflect a careful selection applied by physicians, who opted for requesting the RS test predominantly for N0 patients with more aggressive biological features and for N1 patients with more indolent biology. Focusing the use of the RS test to these challenging clinical situations, referred to a “gray area” of treatment indecision, may uncover, in a more realistic fashion, the impact of the RS test. In addition, patients included in the ROXANE study showed more aggressive biologic features as compared with other large population‐based registries including real‐life data on the use of the RS test. With regard to N0 stage, the rate of patients with grade 3 tumors was about 17% in two population registries, as compared with 49.3% in ROXANE, without relevant differences in T‐stage distribution [18], [19]. In the Surveillance, Epidemiology, and End Results (SEER) registry, N1 patients with grade 1 and 2 breast cancer were 29.0% and 54.8%, respectively, as compared with 9.1% and 73.7% in the ROXANE study. The rate of patients with T >2 cm was similar (32% in the SEER registry and 36% in the ROXANE study) [19].

With these premises, although the pre‐RS indication to HT was high even in the selected ROXANE population (52%), the RS test led to a change in treatment recommendation for 30% of the patients.

In N0 patients, the rate of change was 28%; however, around one third of changes in treatment decisions were from HT to CT + HT. This result is of relevance, because it highlights that the RS test could be useful not only as a tool to spare CT for low‐risk patients but also to avoid undertreatment for high‐risk N0 patients. The ROXANE study was conducted after the results of the TAILORx RS <11 cohort were published [9] but prior to the availability of the full trial results also including the RS 11–25 cohort [10]. Therefore, uncertainty around the most clinically useful cutoff existed at that time. This is highlighted by the lack of reduction in CT use in cases of RS 18–25 in the ROXANE study. A greater impact of the RS test in CT reduction therefore might be hypothesized for N0 patients in light of the recently published full TAILORx data, showing noninferiority of HT versus CT + HT in cases of RS ≤25 [10]. This consideration of an expected larger impact in CT sparing after the TAILORx data may be particularly true for N0 patients older than 50 years, for whom the rate of CT use after RS testing was 55% in ROXANE. On the other hand, TAILORx has shown that younger patients with RS 16–25 may derive some benefit from CT; therefore, the decision of withdrawing CT for these patients may be more complex [10]. Although the distribution of RS results in N0 ROXANE patients was consistent with TAILORx, patients’ clinicopathological characteristics were different between the two studies: grade 3, 49% in ROXANE and 18% in TAILORx; PgR positive, 20% in ROXANE and 10% in TAILORx; modified Adjuvant! Online clinical high risk, 56% in ROXANE and 30% in TAILORx [10]. Therefore, physicians should be aware, when applying the RS test in clinical practice, of the transferability of TAILORx results to the real‐world population.

For N1 patients, the impact of the RS test in the ROXANE study showed a change in treatment decision for up to 33% of the cases, with more than 90% of changes occurring in the direction of sparing CT. This result, together with the low rate of CT use for N1 patients in cases of low RS results (10% for RS ≤11 and 19.5% for RS 11–17), underlines the clinical need of treatment de‐escalation for N1 patients. The results of the ongoing RxPONDER prospective randomized trial, which enrolls patients with HR‐positive/HER2‐negative N1 early breast cancer with RS ≤25 randomized to receive CT + HT or HT, will be available in the next years. Meanwhile, the amount of data from other prospective or large cohort studies for N1 patients is increasing. A recent analysis of 610,350 tumor specimens examined by the Genomic Health laboratory from February 2004 to August 2017 showed that RS result distribution among N0, N1mi, and N1 patients was similar [20]. The prospective West German Study Group PlanB study has reported the outcome of HR‐positive/HER2‐negative N0/N1 patients assigned to adjuvant treatment according to the RS category: RS ≤11, HT; RS >11, randomized to taxane‐based or anthracycline and taxane‐based CT + HT. For N1 patients, the 3‐year disease‐free survival rates were 97.9% for RS ≤11, 97.2% for RS 12–25, and 89.4% for RS >25 [21]. A SEER population‐based study showed the results of more than 10,000 N1 patients with RS results. The 9‐year breast cancer‐specific survival rates were 98.2% for RS ≤11, 99% for RS 11–15, 96.7% for RS 16–20, 93.1% for RS 21–25, and 84.2% for RS ≥26. Despite that the magnitude of the benefit of CT is uncertain in N1 patients as a result of the low number of events, the observed difference in breast cancer‐specific survival with CT in N1 patients was <1% at both 5 years and 9 years [22]. Therefore, while waiting for the data of the large RxPONDER randomized trial, available evidence supports the potential utility of the RS test in N1 patients.

Finally, in our study, we also collected the treatment that was actually received by patients, showing a 5.9% discordance (n = 15 patients) with the post‐RS recommendation, in the vast majority of cases from CT + HT post‐RS to HT alone received. This figure is consistent with previous data from Breast DX, showing that even in a highly selected population there is a low discrepancy from post‐RS recommendation to treatment received [15].

Our study has limitations. Data on clinical outcome were not collected; however, the sample size was too small for this type of analysis. Because of slow accrual, the statistical design was adjusted; however, this did not prevent the observation of the target rate of change in treatment decisions. Potential inherent bias for which the patients were included cannot be excluded; however, the intention of the study was to capture the pattern of use of the RS test in case it would be available for clinical practice. Results in subgroups need to be interpreted with caution because of small sample size. Patient's decisional conflict was not captured.

Conclusion

Although it is difficult to a priori define the population of patients to be tested in order to optimize the use of the RS test, our study shows that clinicians in their routine practice are able to select these patients. When the use of the RS test is left at the clinician's discretion, it is associated with a high rate of change in treatment decision and, importantly, allows sparing chemotherapy for up to 30% of the patients initially recommended to it (reaching up to 50% if only N1 patients are considered), even in a clinical scenario in which most patients with ER‐positive/HER2‐negative early breast cancer are initially recommended to HT alone.

See http://www.TheOncologist.com for supplemental material available online.

Author Contributions

Conception/design: Maria Vittoria Dieci, Valentina Guarneri, Gian Luca De Salvo, PierFranco Conte

Provision of study material or patients: Maria Vittoria Dieci, Valentina Guarneri, Fable Zustovich, Marta Mion, Paolo Morandi, Emilio Bria, Laura Merlini, Pierluigi Bullian, Cristina Oliani, Stefania Gori, Tommaso Giarratano, Enrico Orvieto, Gaia Griguolo, Silvia Michieletto, Tania Saibene, PierFranco Conte

Collection and/or assembly of data: Maria Vittoria Dieci, Valentina Guarneri, Fable Zustovich, Marta Mion, Paolo Morandi, Emilio Bria, Laura Merlini, Pierluigi Bullian, Cristina Oliani, Stefania Gori, Tommaso Giarratano, Gaia Griguolo, Paola Del Bianco, Gian Luca De Salvo, PierFranco Conte

Data analysis and interpretation: Maria Vittoria Dieci, Valentina Guarneri, Paola Del Bianco, Gian Luca De Salvo, PierFranco Conte

Manuscript writing: Maria Vittoria Dieci, Paola Del Bianco, PierFranco Conte

Final approval of manuscript: Maria Vittoria Dieci, Valentina Guarneri, Fable Zustovich, Marta Mion, Paolo Morandi, Emilio Bria, Laura Merlini, Pierluigi Bullian, Cristina Oliani, Stefania Gori, Tommaso Giarratano, Enrico Orvieto, Gaia Griguolo, Silvia Michieletto, Tania Saibene, Paola Del Bianco, Gian Luca De Salvo, PierFranco Conte

Disclosures

Maria Vittoria Dieci: Eli Lilly and Company, Genomic Health, Celgene (C/A); Valentina Guarneri: Eli Lilly and Company, Roche (SAB), Novartis, AstraZeneca (H); Marta Mion: Roche (SAB, C/A); Emilio Bria: Merck Sharp & Dohme, AstraZeneca, Celgene, Pfizer, Helsinn, Eli Lilly and Company, Bristol‐Myers Squibb, Novartis, Roche (SAB), Associazione Italiana Ricerca Cancro, Roche, AstraZeneca (RF), Roche (C/A); PierFranco Conte: Eli Lilly and Company, Novartis, Roche, AstraZeneca (C/A). The other authors indicated no financial relationships.

(C/A) Consulting/advisory relationship; (RF) Research funding; (E) Employment; (ET) Expert testimony; (H) Honoraria received; (OI) Ownership interests; (IP) Intellectual property rights/inventor/patent holder; (SAB) Scientific advisory board