Larotrectinib Compared With Real-World Non–Tropomyosin Receptor Kinase Inhibitor Therapies in Patients With Tropomyosin Receptor Kinase Fusion Cancer

Abstract

PURPOSE

Neurotrophic tyrosine receptor kinase gene fusions are oncogenic drivers of various solid tumors. Larotrectinib is a highly selective tropomyosin receptor kinase (TRK) inhibitor approved for patients with TRK fusion cancer on the basis of single-arm trials. This study was a matched comparative effectiveness study of larotrectinib in clinical trials versus standard of care (SOC) in the real-world (RW) setting.

METHODS

Adult patients with advanced/metastatic TRK fusion non-small cell lung cancer, colorectal cancer, soft tissue sarcoma, thyroid cancer, or salivary gland carcinoma were included. Deduplicated data from RW patients were from US and ex-US data sources. Patients in the larotrectinib cohort (pooled data from three trials, ClinicalTrials.gov identifiers: NCT02122913, NCT02576431, and NCT02637687) were matched 1:1 to RW patients on the basis of tumor type and line of therapy (LOT). A propensity score (weighting) model was used to balance key characteristics between cohorts. The primary outcome was overall survival (OS).

RESULTS

In total, 164 patients were matched 1:1 on tumor type and LOT (82 in each cohort). Balance in the baseline covariates was achieved after weighting. Larotrectinib-treated patients had longer OS (median, not reached [NR] v 37.2 months; hazard ratio [HR], 0.44 [95% CI, 0.23 to 0.83]), time to next treatment (median, NR v 10.6 months; HR, 0.22 [95% CI, 0.13 to 0.38]), duration of therapy (median, 30.8 v 3.4 months; HR, 0.23 [95% CI, 0.15 to 0.33]), and progression-free survival (median, 36.8 v 5.2 months; HR, 0.29 [95% CI, 0.18 to 0.46]) compared with RW patients after propensity score weighting.

CONCLUSION

In TRK fusion cancers, treatment with larotrectinib was associated with longer OS and prolonged time to event compared with SOC in all categories measured. These RW data provide context to support larotrectinib effectiveness in this population.

INTRODUCTION

The tropomyosin receptor kinase (TRK) proteins TRKA, TRKB, and TRKC are encoded by the neurotrophic tyrosine receptor kinase (NTRK) genes NTRK1, NTRK2, and NTRK3, respectively.¹ TRK proteins play a key role in nervous system development and function.¹ NTRK gene fusions are oncogenic drivers of various adult and pediatric cancer types.^1,2 NTRK gene fusions occur with varying frequencies, from up to 90% in rare cancers (eg, secretory breast carcinoma and infantile fibrosarcoma), to <0.5% in more prevalent tumor types (eg, non–small cell lung cancer [NSCLC] and colorectal cancer [CRC]).^3,4 A previous retrospective study conducted between January 2011 and July 2018 evaluating the prognosis of TRK fusion cancer in the real-world (RW) setting found that patients with tumors that harbor NTRK gene fusions do not have a statistically better or worse prognosis than those patients with NTRK wild-type tumors.⁵

CONTEXT

• Key Objective

• How does the effectiveness of larotrectinib compare across the five most common adult tropomyosin receptor kinase (TRK) fusion solid tumor types enrolled in the larotrectinib clinical trial program versus standard of care (SOC) in the real-world (RW) setting?

• Knowledge Generated

• Patients treated with larotrectinib had longer overall survival (median, not reached [NR] v 37.2 months; hazard ratio [HR], 0.44 [95% CI, 0.23 to 0.83]), time to next treatment (median, NR v 10.6 months; HR, 0.22 [95% CI, 0.13 to 0.38]), duration of therapy (median, 30.8 v 3.4 months; HR, 0.23 [95% CI, 0.15 to 0.33]), and progression-free survival (median, 36.8 v 5.2 months; HR, 0.29 [95% CI, 0.18 to 0.46]) compared with RW patients after propensity score weighting.

• Relevance

• The matched comparison with the RW data provides additional evidence to support larotrectinib effectiveness in patients with TRK fusion cancer over non-TRK inhibitor SOC.

Larotrectinib is a first-in-class, highly selective, central nervous system–active TRK inhibitor. In a pooled analysis of three phase I/II single-arm clinical trials, larotrectinib demonstrated robust and durable antitumor efficacy, irrespective of patient age or tumor type, and a favorable safety profile.⁶ This led to the first tumor-agnostic approval of larotrectinib for the treatment of patients with TRK fusion solid tumors in 2018.^7,8 Larotrectinib achieved an investigator-assessed objective response rate (RR) of 79% (95% CI, 72 to 85), a median progression-free survival (PFS) of 28.3 months (95% CI, 22 to not estimable [NE]), and a median overall survival (OS) of 44.4 months (95% CI, 37 to NE) in an expanded data set of 153 evaluable adult and pediatric patients with TRK fusion cancer (data cutoff February 2019).⁹

Conducting a randomized controlled trial in TRK fusion cancers would be very challenging because of overall fusion rarity and cancer type heterogeneity.¹⁰ Therefore, we conducted a matched comparative effectiveness analysis using external RW controls, which used inverse probability of treatment weighting (IPTW). We aimed to retrospectively compare larotrectinib effectiveness across the five most common adult TRK fusion solid tumor types (NSCLC, CRC, thyroid cancer, soft tissue sarcoma [STS], and salivary gland carcinoma) enrolled in the larotrectinib clinical trial program versus standard of care (SOC) in the RW setting.

METHODS

Study Design

This retrospective cohort study compared patients treated with larotrectinib in clinical trials with historical data of patients who received SOC in the RW setting. The study design is shown in Figure 1.

FIG 1.

Study design. ^aFor patients with thyroid cancer, exact matching includes histology (DTC and undifferentiated) and lines of systemic therapies in the advanced/metastatic setting. ^bJuly 2022 data cutoff. ^cRW cohort patients who had received larotrectinib/entrectinib/any investigational drug after index therapy will be censored at the date of first dose of the respective drug. DOT, duration of therapy; DTC, differentiated thyroid cancer; ECOG PS, Eastern Cooperative Oncology Group performance status; IPTW, inverse probability of treatment weighting; OS, overall survival; RW, real-world; rwPFS, real-world progression-free survival; rwRR, real-world response rate; SOC, standard of care; TTNT, time to next treatment.

Study Population

Eligible patients were age 18 years and older at the date of diagnosis of advanced or metastatic TRK fusion NSCLC, CRC, thyroid cancer, STS, or salivary gland carcinoma (these represent the five most common tumor types in the larotrectinib clinical trials). All tumors had NTRK gene fusions, confirmed with next-generation sequencing, reverse transcriptase polymerase chain reaction, polymerase chain reaction, or fluorescence in situ hybridization by local testing.

Patients with a second primary cancer during the study period, except nonmelanoma skin carcinoma or carcinoma in situ, were excluded. Patients involved in larotrectinib clinical trials were excluded from the RW control cohort (those who received TRK inhibitors in the RW setting after index SOC therapy were allowed, but were censored at the initiation of TRK inhibitor therapy).

Data Sources and Study Follow-Up

The larotrectinib clinical trial cohort comprised patient-level data from larotrectinib clinical trials, including a phase I clinical trial in adults age 18 years and older (ClinicalTrials.gov identifier: NCT02122913), a phase II basket clinical trial in adolescents and adults age 12 years and older (NAVIGATE, ClinicalTrials.gov identifier: NCT02576431), and a phase I/II clinical trial in pediatrics younger than 21 years (SCOUT, ClinicalTrials.gov identifier: NCT02637687). The study used the most recent clinical trial data available at the time of the analysis (July 20, 2022) to ensure that all patients had reached the maximum length of follow-up for this study.

In the RW cohort, patients were pooled from four retrospective data sources (the American Association of Cancer Research [AACR] GENIE database,¹¹ Cardinal Health Study,¹² Flatiron Health-Foundation Medicine clinico-genomic database [FH-FMI CGDB],¹³ and ORIEN data source¹⁴) and a global chart review (Data Supplement, Table S1). Detailed information on how these data sources were identified is included in the Data Supplement.

The index date for the comparative analyses was the initiation date of the matched line of therapy (LOT) of larotrectinib or comparator therapy in the advanced or metastatic setting. Patients were followed from index date up to last activity, end of study period, or death, whichever occurred first.

The study end date was aligned with the data collection end date (up to January 2022 for the Cardinal Health Study, December 2022 for FH-FMI CGDB and AACR GENIE, February 2023 for ORIEN, and June 2023 for the global chart review).

Outcomes

The primary outcome was OS (time from index date [initiation of larotrectinib or SOC therapy after matching by tumor type and treatment line] to the time of death [by any cause]). Patients who were still alive were censored at the last known alive date. The RW patients who received a TRK inhibitor were censored at the date of the first dose of the respective drug.

The secondary outcomes included duration of therapy (DOT) and time to next treatment (TTNT). DOT was defined as time from initiation of index therapy until permanent treatment discontinuation, followed by a subsequent therapy, or death for any reason. Patients still on index therapy were censored at the last known assessment date. TTNT was defined as the time from index date until date of initiation of next line of systemic treatment. Patients who did not receive subsequent treatment (ie, they were continuing current treatment at the time of the analysis or lost to follow-up not due to confirmed death) were censored at the date of last known activity (last administration or noncancelled order of a drug contained within the treatment regimen).

The exploratory outcomes were PFS (including real-world PFS [rwPFS]) and RR (including real-world RR [rwRR]); these are described in the Data Supplement.

Matching and Weighting

Patients from larotrectinib trials were matched 1:1 to RW patients on the basis of the tumor type and number of lines of previous systemic therapies to identify the index LOT. For patients with thyroid cancer, matching also included histology (differentiated thyroid cancer [DTC] or non-DTC).

The index LOT of SOC was matched with the larotrectinib treatment line for each tumor type. Matching was done without replacement, so that once a patient was matched, that patient was not eligible for further matching. Matching was conducted in descending order of LOT, to ensure that patients at later LOTs had a chance to be matched. A random selection process was also applied during matching, where patients were randomly selected for a specific LOT.

Propensity scores were generated after identifying matched RW patients to patients from larotrectinib clinical trials. IPTW was generated on the basis of the propensity score. The five variables included in the final propensity score model were sex, race, age, Eastern Cooperative Oncology Group performance status, and presence of metastases at index date. For all baseline covariates included in the final propensity score model, standardized differences between larotrectinib and comparator groups were computed. A standardized difference of ≤10% was considered a reasonable guideline for demonstrating sufficient balance between the two groups.

Analyses

Patient-level data were aggregated across the RW data sources. Data were deduplicated after aggregation using a prespecified two-stage algorithm (Data Supplement).

The study used an exploratory approach and was not designed to test any statistical hypotheses. Demographics and baseline characteristics were summarized with descriptive statistics. P values are descriptive as the study was not designed to demonstrate statistical significance.

All time-to-event end points (OS, DOT, TTNT, and PFS) were analyzed using the Kaplan-Meier method to assess the difference between the larotrectinib clinical trials and RW controls. Time to event between groups was estimated using a Cox proportional hazards model, weighting patients in each group by their inverse probability of treatment.

Sensitivity Analyses for OS

Left truncation analysis used statistical methods to correct for bias introduced by delayed entry of individuals into the study. In the absence of left truncation, patients entered the risk set at the start of the regimen and were followed until they experienced an event or were censored.¹⁵ Left truncation reflects that patients only entered the risk set and began being followed at the time of genomic sequencing.

A 5% trimming of propensity scores was used in the sensitivity analysis. Individuals with the most extreme propensity score values in both groups were removed.

The restricted mean survival times were estimated for patients from larotrectinib clinical trials and RW control patients at various times.

RESULTS

Patient Characteristics

In total, 126 patients from the larotrectinib clinical trials and 144 patients from the RW external comparator cohort across the five predefined cancer types were identified for 1:1 matching (Table 1 and Data Supplement, Table S2). Matching on tumor types and number of lines of systemic therapies in the advanced or metastatic setting generated two cohorts of 82 patients each. There were 28% of patients with STS, 21% with CRC, 18% with NSCLC, 18% with salivary gland carcinoma, and 15% with thyroid cancer (12% with DTC and 2% with non-DTC; Data Supplement, Table S3). There were 48%, 24%, 23%, and 5% of patients in first, second, third, and fourth LOT (as index LOT), respectively.

TABLE 1.

Attrition Table After Design Matching by Tumor Type and Line of Therapy Number

Characteristic	Larotrectinib (n = 82)	RW Controls (n = 82)
Tumor type, No. (%)
CRC	17 (21)	17 (21)
NSCLC	15 (18)	15 (18)
Salivary gland	15 (18)	15 (18)
Soft tissue sarcoma	23 (28)	23 (28)
Thyroid	12 (15)	12 (15)
Index line of therapy number, No. (%)
1	39 (48)	39 (48)
2	20 (24)	20 (24)
3	19 (23)	19 (23)
4	4 (5)	4 (5)

Abbreviations: CRC, colorectal cancer; NSCLC, non–small cell lung cancer; RW, real-world.

Before the cohorts were balanced by weighting, mean age was higher in patients from the RW cohort compared with patients treated with larotrectinib (59 v 53 years). There were also more White patients in the RW cohort than the larotrectinib cohort before weighting (74% v 61%; Table 2). After applying stabilized IPTW to the data set to address potential confounding in comparative analyses between the two treatment groups, the covariates were well balanced, with all standardized differences <10% and overlapping propensity score curves (Data Supplement, Fig S1). The distribution of NTRK fusion partners was similar between larotrectinib and RW cohorts (Data Supplement, Fig S2). Detailed information on index treatments received by matched RW patients is described in the Data Supplement (Table S4).

TABLE 2.

Demographic Characteristics With and Without Stabilized IPTW

Characteristic	Without Stabilized IPTW		With Stabilized IPTW
	Larotrectinib (n = 82)	RW Control (n = 82)	Larotrectinib (n = 82)	RW Control (n = 82)
Age, years
Mean	52.7	58.7	56.3	56.9
Standard deviation	17.2	12.2	16.8	12.1
Race, %
White	61	74	66	67
Other/Unknown	39	26	34	33
Sex, %
Male	48	46	48	48
Female	52	54	52	52
Presence of metastases at index date, %
Yes	93	79	87	86
No	7	21	13	14
ECOG PS at index, %
0-1	87	77	85	83
2-3	13	23	15	17

Abbreviations: ECOG PS, Eastern Cooperative Oncology Group performance status; IPTW, inverse probability of treatment weighting; RW, real-world.

OS

In the unweighted sample, median OS was 63.4 months (95% CI, 42.9 to NE) in patients treated with larotrectinib and 37.2 months (95% CI, 17.4 to NE) in the RW cohort (Fig 2A). In the weighted sample, median OS was not reached (NR) in patients treated with larotrectinib and was 37.2 months (95% CI, 12.5 to NE) in the RW cohort (Fig 2B). Patients treated with larotrectinib had a lower risk of death when compared with patients from the RW cohort in the unweighted (hazard ratio [HR], 0.47 [95% CI, 0.26 to 0.88]; P = .0176) and weighted samples (HR, 0.44 [95% CI, 0.23 to 0.83]; P = .0109).

FIG 2.

OS^a in the (A) unweighted and (B) weighted samples and (C) hazard ratios. ^aIn the RWC, 42 patients were censored because they switched to a TRK inhibitor, investigational drug, or clinical study drug. IPTW, inverse probability of treatment weighting; KM, Kaplan-Meier; NE, not estimable; OS, overall survival; RWC, real-world cohort; TRK, tropomyosin receptor kinase.

The lower risk of death observed in patients treated with larotrectinib was consistent in the sensitivity analysis; this adjusted for lapse between molecular testing and index date¹⁵ (adjusted left truncation [unweighted]: HR, 0.30 [95% CI, 0.16 to 0.56]; P = .0002; adjusted left truncation [weighted]: HR, 0.27 [95% CI, 0.14 to 0.51]; P < .001; IPTW [5% trimming]: HR, 0.46 [95% CI, 0.24 to 0.89]; P = .0202).

The restricted mean survival times were higher in patients treated with larotrectinib than in patients from the RW cohort at 12 months (11.2 months [95% CI, 10.7 to 11.8] v 9.8 months [95% CI, 9 to 10.8]), 24 months (21.4 months [95% CI, 20.1 to 22.8] v 17.7 months [95% CI, 15.5 to 20.2]), and 36 months (30.7 months [95% CI, 28.4 to 33.2] v 24.9 months [95% CI, 21.2 to 29.3]), with corresponding P values of .0134, .0122, and .0224, respectively.

DOT

The median DOT before weighting was longer in patients treated with larotrectinib (29.3 months [95% CI, 15.7 to 38.6]) compared with patients from the RW cohort (3.5 months [95% CI, 2.8 to 4.7]). The HR for DOT was 0.24 (95% CI, 0.17 to 0.36; Fig 3A) before weighting. The median DOT after weighting was longer in patients treated with larotrectinib (30.8 months [95% CI, 18.2 to 47.6]) compared with patients from the RW cohort (3.4 months [95% CI, 2.7 to 4.7]). The HR for DOT was 0.23 (95% CI, 0.15 to 0.33; Fig 3B) after weighting.

FIG 3.

DOT in the (A) unweighted^a and (B) weighted^a samples, and TTNT in the (C) unweighted^b and (D) weighted^b samples. ^aIn the RWC, six patients were censored. ^bIn the RWC, 40 patients were censored. DOT, duration of therapy; NE, not estimable; RWC, real-world cohort; TTNT, time to next treatment.

TTNT

Before weighting, the median TTNT was 58.1 months (95% CI, 53.2 to NE) in patients treated with larotrectinib and 10 months (95% CI, 6.7 to 17.7) in patients from the RW cohort (HR, 0.24 [95% CI, 0.14 to 0.40]; Fig 3C). After weighting, the median TTNT was NR in patients treated with larotrectinib but remained similar in patients from the RW cohort (10.6 months [95% CI, 6.1 to 17.7]; Fig 3D). The HR for TTNT was 0.22 (95% CI, 0.13 to 0.38) after weighting.

Exploratory End Points

PFS defined by all measures before and after weighting was similar (Fig 4). Median PFS was longer in patients treated with larotrectinib (36.8 months [95% CI, 25.8 to 58.2]) compared with rwPFS on the basis of radiologist assessment in patients from the RW cohort (5.2 months [95% CI, 3.5 to 6.8]) after weighting. The HR was 0.29 (95% CI, 0.18 to 0.46; Fig 4B). Consistently, median PFS was also longer in patients treated with larotrectinib (36.8 months [95% CI, 25.8 to 58.2]) compared with rwPFS on the basis of oncologist assessment in patients from the RW cohort (5.9 months [95% CI, 3.7 to 10.2]) after weighting. The HR was 0.33 (95% CI, 0.20 to 0.53; Fig 4D).

FIG 4.

PFS^a and rwPFS^b on the basis of radiologist assessment in the (A) unweighted^c and (B) weighted^c samples, and oncologist assessment in the (C) unweighted^d and (D) weighted^d samples. ^aBased on RECIST v1.1 criteria. ^bDefined as the time from index date to the date of progression (for patients with progression confirmed by imaging or oncologist assessment) or death (within 30 days after last assessment or within 30 days after index line end date). ^cIn the RWC, 22 patients were censored. ^dIn the RWC, 25 patients were censored. PFS, progression-free survival; RW, real-world; RWC, real-world cohort; rwPFS, real-world PFS.

RR is described in the Data Supplement, Table S5.

DISCUSSION

To our knowledge, this is the first comparative effectiveness study using a sequential matching and propensity score approach for a drug approved on the basis of single-arm trials targeting a rare genomic alteration in solid tumors. Our study required a multifaceted global effort to identify RW patients and demonstrates the feasibility of external control-arm studies, even in the context of ultra-rare alterations in precision oncology.¹⁶

Our results demonstrated that all time-to-event end points favored larotrectinib compared with SOC. Patients treated with larotrectinib showed longer OS when compared with an external control group of adult patients with TRK fusion solid tumors who received SOC treatment in the RW setting (median, NR v 37.2 months in the weighted sample). This difference in OS was consistently shown in unweighted, weighted, and sensitivity analyses. DOT and TTNT were also longer for patients treated with larotrectinib compared with patients from the RW cohort. Larotrectinib had a weighted median DOT of 30.8 months and an unweighted median TTNT of 58.1 months (the weighted median TTNT was NR). These outcomes are likely associated with the longer OS of patients treated with larotrectinib. In addition, patients treated with larotrectinib also demonstrated longer PFS compared with patients from the RW cohort. This was shown consistently for both radiologist and medical oncologist assessments. This suggests that the method of assessing progression does not measurably influence the comparison between the patient cohorts. The findings of this study should provide valuable information to help health care providers and patients further understand the benefits of larotrectinib over other non-TRK inhibitor therapeutic options in the treatment of advanced TRK fusion cancers.

This study has several strengths. The study was designed to overcome both the challenges associated with single-arm clinical trials and limitations of indirect adjustment methods (which are not on the basis of adjustment of patient-level data). It included a cohort of RW patients who had confirmed NTRK gene fusions and were treated with current guideline-recommended SOC, which served as an external comparator arm for patients treated with larotrectinib in clinical trials. The study used patient-level data in which the patients from clinical trials were matched to patients with confirmed TRK fusions from the RW. Matching was performed by both tumor type and LOT in the advanced or metastatic setting. In addition, patients with thyroid cancer were further matched by histology.

To allow for exact matching, this study included the five most common adult solid tumor types enrolled in the larotrectinib clinical program⁹; these also represent five of the eight most common TRK fusion adult solid tumor types identified in comprehensive RW analyses.⁴ Using data from five different RW sources maximized sample sizes for this rare disease. Over 40% of patients in the RW cohort were from AACR GENIE sites that are leading international cancer centers, as well as US and international large academic institutions. By incorporating both US and international academic institutions in the control arm, we provided a balanced comparison that reflects RW diversity and reduces potential bias arising from treatment at academic versus community cancer centers. The data sources were evaluated to align with the inclusion criteria of the larotrectinib clinical trial program on the basis of predefined attributes; this ensured the RW data sources served as appropriate external controls. The study applied IPTW via a propensity score model with key variables identified through a systematic literature review. All time-to-event end points provided consistent results.

The current study confirms the benefit of larotrectinib and has certain advantages compared with a previous indirect-matching study in the context of TRK fusion cancer. A previous study from Bokemeyer et al compared treatment outcomes between larotrectinib and SOC using matching-adjusted indirect comparison of RW data from the FH-FMI CGDB.¹⁷ The study did not match patients by tumor type or LOT. The tumor type composition in the two cohorts were different after adjustment and not representative of the most common tumor types with TRK fusions. In addition, the RW cohort in the study had a small number of patients with TRK fusion cancer and SOC was unknown.¹⁷

In the absence of randomized controlled trials, an external control arm on the basis of RW data is invaluable in supplementing single-arm trials for health technology assessment by providing comparative data. It bridges the gap between clinical trial data and RW practice to support more robust and informed decision making for treatments.^18-21

Our study also has limitations. First, there were differences in outcome definitions from RW and clinical trials, since rwPFS and rwRR are not set to measure the same elements as required by RECIST. Second, standardized clinical trial criteria, such as RECIST version 1.1, are not typically captured in medical chart reviews. Third, imaging data may not be widely accessible across RW databases,²² and differences in the frequency of tumor assessments in the RW may introduce bias in comparative-effectiveness studies.²³ Fourth, because of the differences in expected treatment duration among therapies, such as a predefined number of cycles or an indefinite time frame, interpretation of DOT might be complicated. Fifth, the nature of the study design and small sample sizes across the tumor types precluded a meaningful exploration of clinical efficacy of subgroup analyses, such as by tumor type and NTRK gene. Sixth, as the study only included five tumor types, it is possible that these results may not be representative of the general population of patients with less prevalent advanced or metastatic tumors harboring TRK fusions. Finally, given this is an observational study, the possibility of residual confounding cannot be ruled out; as the cohort was only balanced on the basis of variables considered, unmeasured confounding is possible. Outcomes among patients treated in clinical trials may not generalize to the underlying population of patients treated with a disease. Still, it is unlikely that this would completely explain the large effect sizes measured in this analysis despite matching by tumor type, LOT, and propensity scores capturing key prognostic features including age and performance status.

Treatment with larotrectinib was associated with longer OS and all measured time-to-event end points compared with matched patients treated with SOC. These data provide RW evidence supporting the benefit of larotrectinib treatment in patients with TRK fusion cancer over non-TRK inhibitor SOC.

DATA SHARING STATEMENT

Availability of the data underlying this publication will be determined according to Bayer's commitment to the EFPIA/PhRMA “Principles for responsible clinical trial data sharing”. This pertains to scope, time point, and process of data access.

As such, Bayer commits to sharing upon request from qualified scientific and medical researchers patient-level clinical trial data, study-level clinical trial data, and protocols from clinical trials in patients for medicines and indications approved in the United States and the European Union (EU) as necessary for conducting legitimate research. This applies to data on new medicines and indications that have been approved by the EU and US regulatory agencies on or after January 1, 2014.

Interested researchers can use www.vivli.org to request access to anonymized patient-level data and supporting documents from clinical studies to conduct further research that can help advance medical science or improve patient care. Information on the Bayer criteria for listing studies and other relevant information is provided in the member section of the portal.

Data access will be granted to anonymized patient-level data, protocols, and clinical study reports after approval by an independent scientific review panel. Bayer is not involved in the decisions made by the independent review panel. Bayer will take all necessary measures to ensure that patient privacy is safeguarded.

AUTHOR CONTRIBUTIONS

Conception and design: Marcia S. Brose, C. Benedikt Westphalen, Xiaoyun Pan, Vadim Bernard-Gauthier, Milena Kurtinecz, Helen Guo, Neil R. Brett, Vivek Subbiah, Alexander Drilon

Financial support: Xiaoyun Pan

Provision of study materials or patients: Virginie Aris, Vivek Subbiah, Alexander Drilon

Collection and assembly of data: Marcia S. Brose, Xiaoyun Pan, Vadim Bernard-Gauthier, Helen Guo, Neil R. Brett, Abdelali Majdi, Vivek Subbiah, Kenneth L. Kehl

Data analysis and interpretation: All authors

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

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/po/author-center.

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