Vimseltinib versus placebo for tenosynovial giant cell tumour (MOTION): a multicentre, randomised, double-blind, placebo-controlled, phase 3 trial

Summary

Background:

Tenosynovial giant cell tumor (TGCT) is a locally aggressive neoplasm with limited systemic treatment options. This study evaluated the efficacy and safety of vimseltinib, an oral, switch-control, colony-stimulating factor 1 receptor inhibitor, in patients with symptomatic TGCT not amenable to surgery.

Methods:

MOTION is a global, phase 3, double-blind, placebo-controlled trial in 35 centers in 13 countries (ClinicalTrials.gov: NCT05059262; enrollment complete). Eligible patients were at least 18 years old with a histologically confirmed diagnosis of TGCT for which surgical resection would potentially cause worsening functional limitation or severe morbidity. Randomization was 2:1 to vimseltinib (oral 30 mg twice weekly) or placebo using interactive response technology. The primary endpoint was objective response rate (ORR) by independent radiological review using Response Evaluation Criteria in Solid Tumors version 1.1 (RECIST v1.1) at week 25 in the intent-to-treat population. Safety was assessed in patients who received study drug.

Findings:

From January 21, 2022, to February 21, 2023, 123 patients were randomized (vimseltinib, n=83; placebo, n=40). ORR per RECIST v1.1 was significantly higher for vimseltinib versus placebo (40% versus 0%, difference, 40%; 95% confidence interval, 29% to 51%; p<0·0001). Most non-laboratory treatment-emergent adverse events (TEAEs) were grade 1 or 2; the only grade 3 or 4 TEAE in >5% of patients receiving vimseltinib was increased blood creatine phosphokinase (eight patients [10%]). One patient receiving vimseltinib experienced a treatment-related serious TEAE of subcutaneous abscess. There was no evidence of cholestatic hepatotoxicity or drug-induced liver injury.

Interpretation:

Vimseltinib demonstrated a robust ORR and provided statistically significant and clinically meaningful functional and symptomatic improvement. Vimseltinib offers an effective treatment option for patients with TGCT.

Funding:

Deciphera Pharmaceuticals, LLC.

Introduction

Tenosynovial giant cell tumor (TGCT), previously known as pigmented villonodular synovitis or giant cell tumor of the tendon sheath, is a locally aggressive neoplasm affecting the synovium, bursae, and tendon sheath.¹ TGCT is caused by dysregulation of the colony-stimulating factor 1 (CSF1) gene leading to overproduction of CSF1.^1–3 CSF1 overexpression results in tumor growth and expansion by recruiting and inducing local proliferation of CSF1 receptor (CSF1R)-dependent inflammatory macrophages.^4,5

Surgical resection is the standard of care for most patients with TGCT, but not all patients have disease that is amenable to surgery.^1,6 Recurrence rates are approximately 50% for diffuse TGCT, and patients who experience a recurrence are significantly more likely to experience future recurrences.^7–9 Although TGCT is not life-threatening, it is associated with a substantial reduction in quality of life in a patient population whose mean age at diagnosis is 35 to 50 years.¹ These young patients may experience damage requiring total joint replacement; in extreme cases, amputation may be necessary.^1,10 Treatment options for patients with TGCT not amenable to surgery are limited. One systemic agent, pexidartinib, is approved in the US, Taiwan, and Korea, and can only be prescribed under a risk evaluation mitigation strategy in the US due to off-target, rare but potentially fatal cholestatic hepatoxicity.^11–13 Pexidartinib did not gain regulatory approval in Europe due to these safety risks and concerns over the extent and longevity of symptomatic improvements.¹⁴ Patients with TGCT require therapies with manageable toxicity due to the need for long-term treatment. Therefore, an unmet need remains for an effective CSF1R-targeted therapy with a favorable safety profile and demonstrated efficacy that can improve the disabling symptoms of TGCT and overall patient quality of life.

Vimseltinib is an investigational, oral, switch-control tyrosine kinase inhibitor (TKI) specifically designed to selectively and potently inhibit CSF1R.^2,15 In a phase 1/2 study, vimseltinib was well tolerated and demonstrated promising antitumor activity and clinically meaningful changes in patient-reported outcomes (PROs) in patients with TGCT.¹⁶ Here, we compare the efficacy and safety of vimseltinib versus placebo in patients with TGCT not amenable to surgery from the MOTION trial.

Methods

Study design and patients

MOTION is a global, phase 3, double-blind, randomized, placebo-controlled trial in 35 sites in 13 countries (see appendix p 2 for a full list of active sites and investigators). The protocol, protocol amendments, and informed consent documents were approved by an institutional review board or ethics committee at each site and by the appropriate regulatory authorities. This study was performed in accordance with the Declaration of Helsinki and was consistent with International Conference on Harmonisation and Good Clinical Practice guidelines. Applicable local regulatory requirements were followed, and investigators ensured that this study was conducted in full conformity with Regulations for the Protection of Human Subjects of Research.

Patients were at least 18 years old with a histologically confirmed diagnosis of TGCT for which surgical resection would potentially cause worsening functional limitation or severe morbidity, as determined by surgical consultation or multidisciplinary tumor board. Eligible patients had at least one measurable lesion at least 2 cm in size using Response Evaluation Criteria in Solid Tumors version 1.1 (RECIST v1.1), as assessed from magnetic resonance imaging (MRI) by a central radiologist. Patients were excluded if they received previous systemic therapy targeting CFS1 or CSF1R, including vimseltinib; previous therapy with the non-selective inhibitors imatinib and/or nilotinib was allowed. See appendix pp 5–6 for full eligibility criteria. Sex was self-reported (options were “Male” or “Female”), and patients provided written informed consent prior to participation in any study-specific activity.

Randomization and masking

In part one, patients were randomized 2:1 using interactive response technology to receive vimseltinib or matching placebo. Randomization was stratified by tumor location (lower limb versus all other) and region (US versus non-US). A third party that was not involved in the rest of the trial generated the randomization schedule. Patients and site personnel were unblinded to treatment assignment before the end of the double-blind period (week 25) if progressive disease was confirmed by independent radiological review (IRR). At the end of 24 weeks, patients could enter an open-label period (part two, up to week 49); patients who received placebo in part one could cross over to receive vimseltinib in part two. Patients could continue to receive vimseltinib after week 49 in a long-term extension period.

Procedures

Vimseltinib 30 mg twice weekly or matching placebo was administered orally in 28-day cycles for 24 weeks. The study drug was administered each week on day 1 and day 5 (±1 day) at approximately the same time of day on an empty stomach, with at least 24 hours between doses. Dosing could be reduced and/or interrupted at the discretion of the investigator at any time for up to 28 days, or longer if due to an adverse event (AE). If more than two dose reductions were required, the study drug was discontinued. Study drug compliance was assessed by review of dosing diaries and ongoing count of study drug taken.

Patients had two study visits during cycle 1 (day 1 and day 15) and one study visit on day 1 (±7 days) for all subsequent cycles. Clinical laboratory tests and investigator questionnaires (global impression scales) were administered at each visit. Patient questionnaires were administered electronically at all visits after cycle 1 day 1; patients were also asked to complete electronic questionnaires in-between visits. Tumor imaging and range of motion (ROM) assessments were performed at the cycle 4 and cycle 7 (end of part one double-blind period) visits.

Outcomes

The primary endpoint was objective response rate (ORR; complete response + partial response) by IRR using RECIST v1.1 at the beginning of week 25 assessed by MRI (see appendix p 3 for schedule and definition of response).

Key secondary endpoints were also assessed at the beginning of week 25 and were ORR by IRR using tumor volume score (TVS)^17,18; change from baseline in active ROM of the affected joint as measured by goniometry assessments; change from baseline in the following PROs: Patient-Reported Outcomes Measurement Information System physical function (PROMIS-PF; TGCT-specific)^19,20, worst stiffness numeric rating scale (NRS), and EuroQol Visual Analogue Scale (EQ-VAS) health status; and Brief Pain Inventory (BPI) worst pain response rate. BPI worst pain response was defined as at least a 30% improvement without a 30% or greater increase in narcotic analgesic use. Narcotic analgesic use was recorded in a patient diary and reviewed by site staff and the principal investigator at each visit.

Other outcome measures, prespecified in the statistical analysis plan (SAP), reported here are response rates for active ROM, PROMIS-PF, worst stiffness NRS, and EQ-VAS at week 25, which were derived from minimum clinically important difference (MCID) thresholds (change from baseline ≥10%, ≥3, ≤−2, and ≥7, respectively). Change from baseline for BPI worst pain is also reported (prespecified in the SAP). See appendix pp 3–4 for the description of IRR, TVS definition of response, active ROM calculations, and determination of MCID thresholds.

Safety was assessed using the National Cancer Institute Common Terminology Criteria for Adverse Events version 5.0. Adverse event reporting was continuous after receiving informed consent.

Statistical Analyses

The target sample size of 120 patients (vimseltinib, n=80; placebo, n=40) was based on considerations for powering analyses of the primary endpoint, key secondary endpoints, and overall exposure to vimseltinib while assuming a 15% dropout rate. This sample size was planned to have 98% power to detect statistically significant differences between treatment groups, assuming true ORRs of 35% and 5% in the vimseltinib and placebo groups, respectively, using a two-sided Fisher’s exact test at a 5% type 1 error rate.

The intent-to-treat (ITT) population was all patients who were randomized and was the population for all primary and secondary efficacy endpoints. The safety population was patients who received at least one dose of study treatment. Categorical variables were summarized using frequencies and proportions. Time-to-event data were summarized via Kaplan-Meier methodology with two-sided 95% confidence intervals (CIs). The data cutoff was August 22, 2023.

Endpoints were assessed at week 25 and tested hierarchically on the ITT population in the following prespecified order: ORR per RECIST v1.1; ORR per TVS; change from baseline in active ROM, PROMIS-PF, worst stiffness NRS, and EQ-VAS; and BPI worst pain response rate. The p-values noted as nominal were prespecified but not part of the testing hierarchy. ORR per RECIST v1.1, ORR per TVS, and BPI worst pain response were compared between treatment arms using a two-sided Cochran-Mantel-Haenszel test stratified by randomization stratification factors on the ITT population with α=0·05. All other key secondary endpoints were analyzed using a mixed model for repeated measurements (MMRM) to estimate the least squares (LS) mean, with the dependent variable as change from baseline. The fixed effects for each of these models were treatment group, time point, treatment group by time point interaction, stratification factors for tumor location (lower limb versus all other) and region (US versus non-US), and baseline value of the corresponding endpoint. For active ROM only, the stratification factor for tumor location was replaced with affected joint (knee versus ankle versus other).

Missing data were not imputed except for the purpose of identification of treatment-emergent adverse events (TEAEs) and prior or concomitant medications or procedures with missing start or end times. The MMRM treats missing data under the missing at random assumption with implicit imputation. Sensitivity analyses of missing data included multiple imputation based on randomized treatment and a pattern mixture model where patients randomized to vimseltinib who discontinued treatment due to an AE were imputed based on the missing not at random assumption and were imputed as if they received placebo. These results were consistent with the primary method of analysis. For ORR and BPI worst pain, patients who did not have a postbaseline assessment within the week 25 analysis window were considered nonresponders. Data analysis was performed using SAS version 9.4.

A Data Monitoring Committee evaluated safety and efficacy approximately every six months throughout the study. This study is registered with ClinicalTrials.gov, number NCT05059262 and with the European Union Drug Regulating Authorities Clinical Trials Database (EudraCT), number 2020-004883-25.

Role of the funding source

The sponsor, Deciphera Pharmaceuticals, LLC, designed the study with input from the authors. Data analyses were performed by the sponsor according to a prespecified statistical analysis plan, and all authors contributed to results interpretation. The first draft was written by a medical writer, contracted by the sponsor, under direction of the authors; the authors critically reviewed the manuscript and provided substantive input.

Results

From January 21, 2022, to February 21, 2023, 146 adults with TGCT were assessed for eligibility and 123 patients were randomized to vimseltinib 30 mg twice weekly (n=83) or placebo (n=40; figure 1). Overall, nine (11%) patients receiving vimseltinib discontinued treatment in part one (AEs: four [5%]; withdrawal by patient: three [4%]; other: two [2%]). Among patients receiving placebo, five (13%) discontinued treatment in part one (withdrawal by patient: three [8%]; PD per TVS: one [3%]; physician decision: one [3%]). The patient with PD per TVS was eligible to receive vimseltinib in the open-label period of the study (part two).

Figure 1:

Trial profile

*Reasons for screen failures were not captured.

†One patient was prematurely randomized but later deemed not eligible to participate and did not receive placebo.

‡ Both patients in the other category specified lack of clinical benefit as the reason for withdrawal.

§One patient discontinued treatment early in part one and crossed over to receive vimseltinib in the open-label period (part two).

AE=adverse event. IRR=independent radiological review. ITT=intent-to-treat. PD=progressive disease.

Baseline characteristics were balanced between treatment arms (table 1). For patients receiving vimseltinib and placebo, median age (interquartile range [IQR]) was 45 (33–53) and 43 (31–53) years, respectively. In both treatment arms, the majority of patients were White (vimseltinib: 71% [59 of 83 patients]; placebo: 53% [21 of 40 patients]) and/or female (vimseltinib: 55% [46 of 83]; placebo: 68% [27 of 40]). The most common disease location was the knee (vimseltinib: 67% [56 of 83 patients]; placebo: 68% [27 of 40 patients]), and the majority of patients had at least one prior surgery or procedure (vimseltinib: 77% [64 of 83]; placebo: 68% [27 of 40]).

Table 1:

Baseline demographics and clinical characteristics

	Vimseltinib	Placebo
	n=83	n=40
Age, years, median (IQR)	45 (33–53)	43 (31–53)
Sex
Female	46 (55%)	27 (68%)
Male	37 (45%)	13 (33%)
Race
White	59 (71%)	21 (53%)
Asian	1 (1%)	4 (10%)
Black or African American	4 (5%)	0
Other*	19 (23%)	15 (38%)
Region
Europe	57 (69%)	32 (80%)
North America	14 (17%)	5 (13%)
Australia	12 (14%)	2 (5%)
Asia	0	1 (3%)
Affected joint
Knee	56 (67%)	27 (68%)
Ankle	9 (11%)	6 (15%)
Hip	11 (13%)	1 (3%)
Other†	7 (8%)	6 (15%)
Prior TGCT surgery or procedure §	64 (77%)	27 (68%)
1 surgery	28 (34%)	14 (35%)
2–3 surgeries	26 (31%)	11 (28%)
≥4 surgeries	10 (12%)	2 (5%)
Prior TGCT systemic therapy	19 (23%)	9 (23%)‡
Imatinib	16 (19%)	7 (18%)
Nilotinib	2 (2%)	4 (10%)
Other¶	1 (1%)	0
Size of target lesions, mm, median (IQR)	63.4 (36·2, 88·7)	61.0 (44·0, 85·2)
Narcotic analgesic use	12 (14%)	5 (13%)

Data shown as n (%) unless otherwise noted. ITT population includes patients who were randomized to receive study drug.

^*Includes not reported and unknown.

^†Includes foot (vimseltinib: n=1, placebo: n=3), wrist (vimseltinib: n=2, placebo n=1), hand (vimseltinib: n=2, placebo: n=0), shoulder (vimseltinib: n=1, placebo: n=1), elbow (vimseltinib: n=1, placebo: n=0), and temporomandibular joint (vimseltinib: n=0, placebo: n=1).

^‡Two patients in the placebo arm received both imatinib and nilotinib.

^§Diagnostic biopsies were not recorded as a prior surgery or procedure.

^¶Includes an investigational agent.

ITT=intent-to-treat. IQR=interquartile range. TGCT=tenosynovial giant cell tumor.

The primary endpoint of ORR by IRR per RECIST v1.1 at week 25 was 40% (33 of 83 patients) for vimseltinib versus 0% (zero of 40 patients) for placebo (difference, 40%; 95% CI, 29% to 51%; p<0·0001; table 2). Responses with vimseltinib included four (5%) patients with complete response and 29 (35%) with partial response. The ORR at week 25 in subgroups defined by baseline characteristics, including patient sex, all favored treatment with vimseltinib (figure 2, appendix p 10). Additionally, nearly all patients receiving vimseltinib experienced a decrease in tumor size per RECIST v1.1 (figure 3). The median DOR for responders at week 25 was not reached at data cutoff (range, 0·03+ to 11·7+ months).

Table 2:

Primary and key secondary endpoints at week 25 in patients with TGCT

	Vimseltinib	Placebo
	n=83	n=40
Overall response by IRR using RECIST v1.1
Complete response	4 (5%)	0
Partial response	29 (35%)	0
Stable disease	42 (51%)	33 (83%)
Not evaluable	8 (10%)	7 (18%)
ORR by IRR using RECIST v1.1	33 (40%)	0
Difference, % (95% CI), p-value*	40% (29 to 51), p<0·0001	··
DOR by IRR using RECIST v1.1, months, median† (min, max)	NR (0·03+, 11·7+)	N/A
Overall response by IRR using TVS
Complete response	4 (5%)	0
Partial response	52 (63%)	0
Stable disease	19 (23%)	34 (85%)
Progressive disease	0	1 (3%)
Not evaluable	8 (10%)	5 (13%)
ORR by IRR using TVS	56 (67%)	0
Difference, % (95% CI), p-value	67% (56 to 77), p<0·0001	··
DOR by IRR using TVS, months, median† (min, max)	NR (0·03+, 13·9+)	N/A
Active ROM
Baseline, mean (SD)	63·0% (29·4)	62·9% (32·2)
Week 25, mean (SD)	83·6% (28·1)	68·3% (35·3)
Change from baseline, %, LS mean (SE)	18.4% (6·5)	3.8% (7·2)
Difference, % (95% CI), p-value	14·6% (4·0 to 25·3), p=0·0077	··
Responder rate‡, n (%)	40 (48%)	8 (20%)
PROMIS-PF
Baseline, mean (SD)	39·0 (6·1)	38·5 (6·0)
Week 25, mean (SD)	43·7 (6·1)	40·7 (6·7)
Change from baseline, LS mean (SE)	4·6 (1·0)	1·3 (0·9)
Difference (95% CI), p-value	3·3 (1·4 to 5·2), p=0·0007	··
Responder rate§, n (%)	36 (43%)	10 (25%)
Worst stiffness NRS
Baseline, mean (SD)	5·1 (2·0)	5·2 (1·8)
Week 25, mean (SD)	2·9 (2·1)	4·3 (1·9)
Change from baseline, LS mean (SE)	−2·1 (0·2)	−0·3 (0·3)
Difference (95% CI), p-value	−1·8 (−2·5 to −1·1), p<0·0001	··
Responder rate¶, n (%)	32 (39%)	6 (15%)
EQ-VAS
Baseline, mean (SD)	61·4 (19·5)	60·2 (20·6)
Week 25, mean (SD)	74·1 (15·0)	67·5 (15·9)
Change from baseline, LS mean (SE)	13·5 (2·4)	6·1 (2·9)
Difference (95% CI), p-value	7·4 (1·4 to 13·4), p=0·016	··
Responder rate**, n (%)	31 (37%)	10 (25%)
BPI worst pain
Responder rate††, n (%)	40 (48%)	9 (23%)
Difference, % (95% CI), p-value*	26% (4 to 42), p=0·0056	··

Data shown as n (%) unless otherwise noted. Patients who did not have an assessment at the end of part one for any reason or whose week 25 assessment was after the first dose in the open-label period or outside of the visit window of ±14 days were assessed as not evaluable and a nonresponder. ITT population includes patients who were randomized to receive study drug.

^*An unstratified exact CI was utilized.

^†Based on Kaplan-Meier estimate. DOR is defined as time from first imaging result showing response to disease progression or death by any cause.

^‡Responder: Change from baseline in active ROM at week 25 was at least 10%.

^§Responder: Change from baseline in PROMIS-PF at week 25 was at least 3.

^¶Responder: Change from baseline in worst stiffness at week 25 was equal to or less than −2.

^**Responder: Change from baseline in EQ-VAS at week 25 was at least 7.

^††Responder: Experienced at least a 30% decrease in mean BPI worst pain and did not experience a 30% or greater increase in narcotic analgesic use.

BPI=Brief Pain Inventory. CI=confidence interval. DOR=duration of response. EQ-VAS=EuroQol Visual Analogue Scale. IRR=independent radiological review. ITT=intent-to-treat. LS=least squares. NRS=numeric rating scale. min=minimum. max=maximum. N/A=not applicable. NR=not reached. ORR=objective response rate. PROMIS-PF=Patient-Reported Outcomes Measurement Information System physical function. RECIST v1.1=Response Evaluation Criteria in Solid Tumors version 1.1. ROM=range of motion. SD=standard deviation. SE=standard error. TGCT=tenosynovial giant cell tumor. TVS=tumor volume score.

Figure 2:

Forest plot of objective response by patient subgroup

*Large joints were shoulder, elbow, hip, or knee.

†Small joints were all other joints.

An unstratified Wald CI is provided.

CI=confidence interval. TGCT=tenosynovial giant cell tumor.

Figure 3:

Best percent change in target lesions as assessed by IRR using RECIST v1.1

The dotted line at 20% represents threshold for PD; the dotted line at −30% represents the threshold for PR. Shows individual patient values for those with assessment at the end of part one.

CR=complete response. IRR=independent radiological review. NE=not evaluable. PD=progressive disease. PR=partial response. RECIST v1.1=Response Evaluation Criteria in Solid Tumors version 1.1. SD=stable disease.

A key secondary endpoint, ORR by IRR per TVS at week 25, was significantly higher for vimseltinib versus placebo (vimseltinib: 67% [56 of 83 patients]; placebo: 0% [zero of 40 patients]; difference, 67%; 95% CI, 56% to 77%; p<0·0001; table 2), including four (5%) patients with complete response and 52 (63%) with partial response. The LS mean change from baseline in active ROM of the affected joint at week 25 was significantly higher with vimseltinib versus placebo (18·4% versus 3·8%, respectively; difference, 14·6 percentage points; 95% CI, 4·0 to 25·3; p=0·0077; table 2). The active ROM response rate was significantly higher in patients receiving vimseltinib versus placebo (48% [40 of 83 patients] versus 20% [8 of 40 patients], respectively; difference, 28%; 95% CI, 12% to 45%; nominal p=0·0025). Improvements in ROM were not limited to patients with an objective tumor response per RECIST v1.1; patients with stable disease also showed improvements in active ROM (appendix p 11). All six key secondary PRO measures demonstrated clinically meaningful differences for patients receiving vimseltinib versus placebo (table 2). Vimseltinib significantly improved physical function with an LS mean change from baseline in PROMIS-PF of 4·6 for patients receiving vimseltinib versus 1·3 for those receiving placebo (difference, 3·3; 95% CI, 1·4 to 5·2; p=0·0007), and PROMIS-PF response rates were 43% (36 of 83 patients) and 25% (10 of 40 patients) for vimseltinib versus placebo, respectively (difference, 18%; 95% CI, 1% to 36%; nominal p=0·046). Additionally, patients receiving vimseltinib experienced statistically significant improvements in stiffness with LS mean change from baseline in worst stiffness NRS of −2·1 and −0·3 for vimseltinib and placebo, respectively (difference, −1·8; 95% CI, −2·5 to −1·1; p<0·0001), and worst stiffness NRS response rates were 39% (32 of 83 patients) with vimseltinib and 15% (6 of 40 patients) with placebo (difference, 24%; 95% CI, 8% to 39%; nominal p=0·0080). The LS mean change from baseline for EQ-VAS was significantly higher with vimseltinib versus placebo (13·5 versus 6·1, respectively; difference, 7·4; 95% CI, 1·4 to 13·4; p=0·016); 37% (31 of 83 patients) of patients receiving vimseltinib were EQ-VAS responders versus 25% (10 of 40 patients) receiving placebo (difference, 12%; 95% CI, −5% to 29%; nominal p=0·16). The BPI worst pain response rate for patients receiving vimseltinib was 48% (40 of 83 patients) versus 23% (9 of 40 patients) for placebo (difference, 26%; 95%, CI 4% to 42%; p=0·0056). Completion rates for all PRO assessments were robust and consistent at week 25, and there were no substantial differences in the rates of missing data between treatment arms (appendix pp 12–15). In a separate sensitivity analysis, if the missing data for all patients were imputed based on the mean value using multiple imputation for patients that received placebo, then the differences in LS mean change from baseline at week 25 remained significant for all measures (appendix p 7). Vimseltinib provided early and durable improvements versus placebo in mean change from baseline over time for active ROM and all PROs (appendix p 16). Symptomatic improvements were observed in patients receiving vimseltinib who had either an objective tumor response or stable disease per RECIST v1.1, and more patients experienced improvement in multiple measures with vimseltinib versus placebo. Clinically meaningful improvements in three or more measures of functional health or symptoms were observed in 42% of patients receiving vimseltinib who achieved complete response or partial response and 43% who achieved stable disease, compared with only 6% of patients receiving placebo who had stable disease (appendix p 16).

In the safety population, 39 of 40 patients received placebo; one ineligible patient was prematurely randomized to placebo and did not receive treatment. Grade 3 or 4 TEAEs occurred in 37% (31 of 83 patients) and 10% (four of 39 patients) of patients receiving vimseltinib and placebo, respectively (appendix p 8). There was one treatment-related serious TEAE in a patient receiving vimseltinib (subcutaneous abscess; the patient had a medical history of abscess) and none with placebo. Treatment-related grade 3 or 4 TEAEs were reported in 30% (25 of 83 patients) and 3% (one of 39 patients) of patients receiving vimseltinib and placebo, respectively. There were no deaths due to TEAEs in either treatment arm. The majority of non-laboratory TEAEs were grade 1 or 2 for both arms (table 3). The most common TEAEs (>20%) in patients receiving vimseltinib were periorbital edema (45% [37 of 83 patients]), fatigue (33% [27 of 83]), face edema (31% [26 of 83]), pruritus (29% [24 of 83]), headache (28% [23 of 83]), asthenia (27% [22 of 83]), nausea (25% [21 of 83]), increased blood creatine phosphokinase (24% [20 of 83]), and increased aspartate aminotransferase (23% [19 of 83]). The most common TEAEs with placebo were headache (26% [ten of 39 patients]), asthenia (23% [nine of 39]), diarrhea (21% [eight of 39]), and nausea (21% [eight of 39]). The only grade 3 or 4 TEAE in >5% of patients receiving vimseltinib was increased blood creatine phosphokinase; this was not associated with skeletal muscle injury or other organ damage. The observed elevations of serum enzymes were consistent with the known effect of Kupffer cell inhibition of enzyme clearance.^21,22 There was no evidence of cholestatic hepatotoxicity or drug-induced liver injury (appendix p 9). There were no grade 3 or 4 TEAEs in >5% of patients receiving placebo. TEAEs led to dose reduction, dose interruption, and treatment discontinuation in 42% (35 of 83 patients), 53% (44 of 83), and 6% (five of 83), respectively, of patients receiving vimseltinib. The median (IQR) number of interruptions per patient and number of doses missed per interruption for patients receiving vimseltinib were 1 (0–1) and 4 (2–6)), respectively. The median (IQR) time spent in part one on the standard dose (30 mg) of vimseltinib was 23·6 weeks (16·0–23·7). Among patients who experienced dose reductions, the median (IQR) time spent in part one on the first (20 mg) and second (14 mg) dose reduction of vimseltinib was 6·9 weeks (4·6–11·7) and 6·6 weeks (5·7–7·3), respectively. The majority of patients (70% [58 of 83 patients]) were receiving the standard 30-mg twice-weekly dose at the end of the double-blind period.

Table 3:

TEAEs in at least 10% of patients with TGCT in either treatment arm

	Vimseltinib		Placebo
	n=83		n=39*
Preferred term	All grades	Grade 3/4	All grades	Grade 3/4
Periorbital edema	37 (45%)	3 (4%)	5 (13%)	0
Fatigue	27 (33%)	0	6 (15%)	0
Face edema	26 (31%)	1 (1%)	3 (8%)	0
Pruritus	24 (29%)	2 (2%)	3 (8%)	0
Headache	23 (28%)	1 (1%)	10 (26%)	0
Asthenia	22 (27%)	1 (1%)	9 (23%)	1 (3%)
Nausea	21 (25%)	0	8 (21%)	1 (3%)
Blood CPK increased	20 (24%)	8 (10%)	0	0
AST increased	19 (23%)	0	1 (3%)	0
Arthralgia	16 (19%)	0	6 (15%)	1 (3%)
Rash	16 (19%)	0	2 (5%)	0
Rash maculopapular	16 (19%)	1 (1%)	0	0
Edema peripheral	15 (18%)	0	3 (8%)	0
Hypertension	14 (17%)	4 (5%)	4 (10%)	1 (3%)
Generalized edema	11 (13%)	1 (1%)	0	0
Eyelid edema	11 (13%)	0	2 (5%)	0
Lacrimation increased	10 (12%)	0	0	0
Diarrhea	10 (12%)	0	8 (21%)	1 (3%)
COVID-19	10 (12%)	1 (1%)	1 (3%)	0
ALT increased	9 (11%)	0	1 (3%)	0

Data are shown as n (%). AE severity was documented using the National Cancer Institute Common Terminology Criteria for Adverse Events version 5.0. No grade 5 events (deaths) occurred during the trial.

^*One patient randomized to placebo never received treatment.

AE=adverse event. ALT=alanine aminotransferase. AST=aspartate aminotransferase. COVID-19=coronavirus disease 2019. CPK=creatine phosphokinase. TEAE=treatment-emergent AE. TGCT=tenosynovial giant cell tumor.

Discussion

There are limited systemic treatment options with demonstrated efficacy for patients with symptomatic TGCT whose disease is not amenable to surgery, posing a significant unmet medical need for these patients. The phase 3, randomized, placebo-controlled MOTION trial evaluated the efficacy and safety of vimseltinib in this population. Treatment with vimseltinib provided statistically significant and clinically meaningful improvements versus placebo for the primary and all six key secondary endpoints.

The antitumor effect of vimseltinib was robust, including four (5%) patients who experienced complete response. For patients receiving vimseltinib, ORR at week 25 using RECIST v1.1 and TVS was 40% (33 of 83 patients) and 67% (56 of 83), respectively. The response observed with pexidartinib in the ENLIVEN phase 3 trial in which the ORRs by RECIST v1.1 and TVS at week 25 were 39% (24 of 61 patients) and 56% (34 of 61), respectively.¹⁸ While not directly comparable, patients treated with vimseltinib in MOTION had a substantially higher ORR at week 25 than the best overall ORR on study in patients treated with the less-CSF1R-selective and unapproved TKIs imatinib (31% [17 of 55 patients]) and nilotinib (6% [three of 51 patients]), in a retrospective analysis and a phase 2 trial, respectively.^23,24 The ORRs by RECIST v1.1 and TVS at week 25 in the MOTION phase 3 study confirm the results of the ongoing phase 2 study of vimseltinib in patients with TGCT who did not receive prior anti-CSF1/CSF1R therapy.²⁵

TGCT is characterized by severe functional limitations and substantial morbidity, leading to a disabling impact on daily activities.^1,26 With a mean age at diagnosis of 35 to 50 years,¹ long-term quality of life should be a priority when evaluating treatment options in these young patients. Consistent with the epidemiology of TGCT, the median (IQR) age for all patients in MOTION was 44 (32–53) years. Vimseltinib provided statistically significant and clinically meaningful improvements in active ROM, physical function, stiffness, health status, and pain versus placebo. The PRO results reported here are high quality due to the robust collection and high completion rates for all PRO measures. In the ENLIVEN trial, there were high numbers of missing observations across secondary efficacy measures (ROM and PROs), leading the European Medical Agency to conclude that there was not sufficient evidence to show an association between pexidartinib antitumor activity and functional or symptomatic improvements.¹⁴ Clinically meaningful improvements in endpoints such as active ROM, stiffness, or pain may translate to impactful differences in quality of life, such as being able to bathe independently, climb stairs without assistance, or perform tasks for employment.²⁶ Additionally, in MOTION, symptomatic benefit was not limited to patients who had objective responses per RECIST v1.1; patients receiving vimseltinib who had stable disease also experienced meaningful improvement in active ROM and PRO measures. These patients likely had reductions in tumor size that did not reach the RECIST threshold for partial response (≥30% reduction), indicating that even small reductions in lesion size can have a large impact on a patient’s functional health. Furthermore, many disease symptoms may be caused by the inflammatory nature of TGCT and not necessarily the tumor mass itself.²⁷ Vimseltinib can dramatically reduce these symptoms by blocking CSF1 signaling between neoplastic and inflammatory cells without achieving objective response. Additionally, it is widely documented that measuring the size of TGCT using linear methods, such as those employed by RECIST, can be difficult due to irregular tumor shape, complex morphology, and poor imaging contrast in some locations.^17,28,29 Vimseltinib demonstrated robust response according to RECIST, but some patients with stable disease per RECIST were responders per TVS criteria, which may explain the symptomatic benefit reported by patients without a response by RECIST definitions.²⁹

Vimseltinib was well tolerated, with the majority of non-laboratory TEAEs being grade 1 or 2. The only grade 3 or 4 TEAE in >5% of patients treated with vimseltinib was increased blood creatine phosphokinase, which is consistent with the known mechanism of action of CSF1R inhibitors and not felt to be clinically relevant.^21,22 Dose interruptions occurred in approximately half of patients (43 of 83 patients) receiving vimseltinib but were short in duration (median [IQR] number of doses missed per interruption: 4 [2–6]), and the median (IQR) number of dose interruptions per patient was 1 (0–1). Although dose reductions occurred on study, most patients were receiving the 30-mg twice-weekly dose at the end of the double-blind period (70% [58 of 83 patients]), and the rate of discontinuation due to AEs remained low (6% [5 of 83]). A primary concern with pexidartinib is the off-target, rare and unpredictable risk of serious cholestatic or mixed liver injury, and therefore, patients receiving this drug require more frequent clinical and laboratory monitoring; however, in MOTION, there was no evidence of cholestatic hepatotoxicity or drug-induced liver injury with vimseltinib.^11,12,18,30 The observed hepatotoxicity with pexidartinib is considered an off-target effect and not due to CSF1R inhibition.¹⁴ Additionally, cholestatic hepatotoxicity has not been reported for other small molecules or antibodies that inhibit CSF1R.⁴ Taken together with the safety profile of vimseltinib and other CSF1R inhibitors, these findings suggest that the liver injury associated with pexidartinib is not a class effect of CSF1R inhibitors, but rather specific to the drug itself. Furthermore, pexidartinib is a multikinase inhibitor whereas vimseltinib is specific for CSF1R, which may further explain the lack of off-target effects.^2,29 This multikinase activity is evident by the hair hypopigmentation experienced by some patients receiving pexidartinib, which can be attributed to inhibitory effects on KIT;^14,18 this AE was not observed in patients receiving vimseltinib.

In the phase 1 portion of the ongoing phase 1/2 trial of vimseltinib, the median treatment duration is 25·1 months (IQR, 9·4–33·4), with approximately 4 years being the longest time on treatment.³¹ The optimal treatment duration with anti-CSF1R agents for patients with TGCT remains to be determined. The MOTION study is ongoing and results from the open-label period (part two) and long-term extension are forthcoming to support the continued efficacy and safety of vimseltinib. The use of a placebo during the double-blind period (part one) was necessary in MOTION because there is no globally accepted standard of care active comparator. Additionally, patients receiving placebo in part one were able to cross over to receive vimseltinib in the open-label period (part two; 35 of 40 patients [88%] crossed over). Based on the robust activity of vimseltinib observed in part one, similar response rates are expected in the patients that cross over in the open-label period (part two). Responses are also expected to deepen over time, similar to results from the phase 2 portion of the ongoing phase 1/2 study, in which patients who did not receive prior anti-CSF1/CSF1R therapy had a best overall ORR of 64% (29 of 45 patients) with a median treatment duration (IQR) of 21·0 months (7·3–23·9).²⁵

Limitations of the MOTION study include those inherent to all randomized controlled trials, such as difficulty generalizing the findings to the wider public. Since TGCT is a rare disease, many study sites are needed to reach target enrollment, which potentially introduces referral bias and difference in treatment patterns. Enrollment was statistically powered for the primary endpoint, however, the low numbers of patients in the subgroup analyses limit the strength of conclusions from those results. The heterogeneity of TGCT also serves as a limitation because quality of life and PROs may be more impacted by tumors in some anatomic locations versus others. Patients in the placebo arm received best supportive care; therefore, patients receiving placebo in MOTION may demonstrate greater improvements than a completely untreated patient population. Additionally, MOTION did not include patients who previously received anti-CSF1/CSF1R therapy (excluding imatinib or nilotinib), which likely contributed to the larger number of patients from outside the United States where pexidartinib is not approved. However, preliminary phase 1/2 results indicate that vimseltinib is effective in patients who previously received anti-CSF1/CSF1R therapy, including pexidartinib.³² Another limitation is that the irregular shape and asymmetrical growth of TGCT makes measuring changes using RECIST v1.1 challenging.³³

Overall, vimseltinib demonstrated significant antitumor activity versus placebo, with a favorable safety profile in patients with TGCT not amenable to surgery. Patients experienced statistically significant and clinically meaningful improvement in active ROM, which could provide relief from mobility-related limitations in this young population. Finally, patients treated with vimseltinib reported significantly improved stiffness, physical function, health status, and pain. If approved, vimseltinib offers an effective systemic treatment option to patients with TGCT and provides functional health and symptomatic benefit to a population living with substantial morbidity and limited treatment options.

Research in context.

Evidence before this study

Treatment options for patients with tenosynovial giant cell tumor (TGCT) whose disease is not amenable to surgery are limited. Pexidartinib is the only colony-stimulating factor 1 receptor (CSF1R) inhibitor approved in the US, but it is associated with the rare and unpredictable risk of serious cholestatic or mixed liver injury. We searched PubMed for original articles and reviews published between January 1, 2000, and October 1, 2023, on the use of tyrosine kinase inhibitors in TGCT. The search terms used were “TGCT”, “tenosynovial giant cell tumor”, “PVNS”, “tyrosine kinase inhibitor”, “CSF1R”, “clinical trial”, and “vimseltinib”. We identified limited clinical studies using CSF1R inhibitors in TGCT, especially those that include both functional and symptomatic outcomes.

Added value of this study

MOTION is the first global, phase 3, double-blind, randomized trial of vimseltinib, an investigational, oral, switch-control, tyrosine kinase inhibitor designed to selectively and potently inhibit CSF1R, in patients with TGCT whose disease is not amenable to surgery. Patients treated with vimseltinib experienced statistically significant and clinically meaningful improvements in the primary endpoint and all six key secondary endpoints versus placebo. Objective response rate by Response Evaluation Criteria in Solid Tumors version 1.1 and by tumor volume score with vimseltinib were 40% and 67%, respectively. Active range of motion, physical function, stiffness, health status, and pain were significantly improved for vimseltinib versus placebo. Most non-laboratory treatment-emergent adverse events with vimseltinib were grade 1 or 2, and there was no evidence of cholestatic hepatotoxicity or drug-induced liver injury.

Implications of all the available evidence

Patients with TGCT require therapies with manageable toxicity due to the need for long-term treatment. Vimseltinib may address the significant unmet medical needs in TGCT and offers an effective potential systemic treatment option to patients that provides functional health and symptomatic benefit to a population living with substantial morbidity and limited treatment options.

Vimseltinib has the potential to become the standard of care systemic therapy in TGCT based on these very positive results.

Data sharing statement

Qualified scientific and medical researchers can make requests for individual participant data that underlie the results reported in this article, after de-identification, at info@deciphera.com. Proposals for data will be evaluated and approved by Deciphera Pharmaceuticals, LLC at its sole discretion. All approved researchers must sign a data access agreement before accessing the data. Data will be available as soon as possible but no later than within 1 year of the acceptance of the article for publication and for 3 years after article publication. Deciphera Pharmaceuticals, LLC will not share data from identified participants or a data dictionary.