A phase 4, multicenter, global clinical study to evaluate discontinuation and rechallenge of pexidartinib in patients with tenosynovial giant cell tumor previously treated with pexidartinib

Abstract

Background

Pexidartinib is effective in patients with tenosynovial giant cell tumor (TGCT) for whom surgery is not feasible. Durability of response after discontinuation of pexidartinib and the safety and efficacy of restarting pexidartinib have not been previously recorded. This phase 4 study was designed to mimic the real‐world experience with pexidartinib to evaluate the effects of discontinuation of and retreatment with pexidartinib in patients with TGCT who previously benefited from the drug.

Methods

This was a global, multicenter, phase 4 study that enrolled patients with TGCT who were experiencing clinical benefit from pexidartinib in one of four prior phase 1 or phase 3 studies investigating pexidartinib in the disease. Patients could choose to continue pexidartinib at the same dose (the treatment‐continuation cohort) or discontinue treatment with the option to restart pexidartinib (the treatment‐free/retreatment cohort). Tumor progression determined by Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1, patient‐reported outcomes (the Patient‐Reported Outcomes Measurement Information System–Physical Function [PROMIS‐PF] questionnaire and the EuroQol 5‐dimension, 5‐level [EQ‐5D‐5L] visual analog scale), and safety were assessed every 3 months. The primary end point was the proportion of patients in the treatment‐free/retreatment cohort who remained treatment‐free at month 12 and 24; this did not depend on disease progression.

Results

Thirty‐two patients were enrolled: 21 chose to enter the treatment‐continuation cohort, and 11 entered the treatment‐free/retreatment cohort. During the treatment‐free period, six of 11 (54.5%) patients in the treatment‐free/retreatment cohort had progressive disease (PD) according to RECIST, version 1.1, whereas no patient in the treatment‐continuation cohort had disease progression. Over the 24‐month study, three of 11 (27.3%) patients in the treatment‐free/retreatment cohort restarted treatment because of RECIST version 1.1 PD, symptomatic progression, or both (n = 1 each). The probability of remaining treatment‐free in the treatment‐free/retreatment cohort was 73% (95% confidence interval, 37%–90%). In the treatment‐free/retreatment cohort, the median progression‐free survival of the treatment‐free period was 22.8 months (95% confidence interval, 1.6 months to not estimable). By 6 months of retreatment, all retreated patients achieved new disease stabilization with no new safety concerns; two patients had clinically significant improvements in PROMIS‐PF and EQ‐5D‐5L visual analog scale scores. The mean PROMIS‐PF and EQ‐5D‐5L scores remained stable throughout the study. There was no hepatotoxicity and no new safety signal in either cohort.

Conclusions

In this small phase 4 study designed to evaluate outcomes in patients who stopped and restarted pexidartinib, 54.5% of patients who discontinued pexidartinib showed PD, with a median progression‐free survival of 22.8 months. Each of the three patients who restarted pexidartinib stopped progressing, and some reported a new gain in physical function. No PD was detected in patients who remained on treatment, and the safety profile did not indicate long‐term hepatotoxicity or any new safety concerns.

Plain Language Summary

• Pexidartinib is a medication for people with tenosynovial giant cell tumor (TGCT) for whom surgery is not recommended.

• This study examined what happens when patients stop taking pexidartinib (after it has helped them) to see whether it is safe and effective to restart if needed.

• Thirty‐two patients were followed for 2 years: 21 chose to continue pexidartinib and did not experience disease progression, while 11 chose to stop pexidartinib; 54.5% of them had disease progression within 2 years.

• Three patients who stopped pexidartinib restarted treatment after their tumor size increased or their symptoms worsened.

• Restarting pexidartinib was safe and effective in these patients.

INTRODUCTION

Tenosynovial giant cell tumor (TGCT) is a rare, nonmalignant mesenchymal tumor of the synovium, bursae, or tendon sheaths that is driven by overexpression of colony‐stimulating factor‐1 (CSF‐1). ¹ , ² , ³ Surgery is the primary treatment for patients with TGCT when complete tumor resection is feasible; however, given the high frequency of disease recurrence, particularly in the diffuse subtype, patients may benefit from systemic therapy when surgical options are limited. ² , ⁴ Pexidartinib is an oral CSF‐1 receptor (CSF1R) inhibitor approved in the United States, Taiwan, and South Korea for adult patients who have symptomatic TGCT associated with severe morbidity or functional limitations that is not amenable to improvement with surgery (or other treatment; Taiwan only). ⁵ , ⁶ , ⁷ Approval was based on the phase 3 ENLIVEN study (ClinicalTrials.gov identifier NCT02371369) that showed a Response Evaluation Criteria in Solid Tumors, version 1.1 (RECIST v1.1) overall response rate (ORR) of 39% at week 25, which was the primary end point of the study. ⁸

Pexidartinib is available through a Risk Evaluation and Mitigation Strategy (REMS) program because of the risk of potentially fatal hepatotoxicity. ⁵ Further clinical studies demonstrated that pexidartinib continued to be effective over a median of 39 months, with a RECIST v1.1 ORR of 60%. ⁹ In the ENLIVEN study, no cases of rapid tumor regrowth were observed within the 90‐day post‐treatment observation period. ⁸ Beyond the 3‐month post‐treatment follow‐up in ENLIVEN, there are no data to fully understand the impact of discontinuing pexidartinib treatment in patients with stable and/or residual disease. Overexpression of CSF‐1 by a minority of cells is the cause of TGCT, and this attracts reactive, nonneoplastic cells—primarily macrophages—to the joint, causing the tumorous mass. ¹ It is unknown how pexidartinib affects the CSF‐1–overexpressing cells, but it is likely that the tumorigenic clone remains, leading to eventual recurrence after treatment discontinuation. Thus the feasibility of a drug holiday remains an open medical question.

Data from real‐world studies report the average age at diagnosis of TGCT is 39–46 years, suggesting that the need for systemic treatment could last for decades. ¹⁰ , ¹¹ Health care providers have reported that patients may wish to discontinue pexidartinib treatment for reasons like family planning, adverse events (AEs; such as hypopigmentation, rash, or edema), or work assignments. Patients may also experience a dramatic tumor response and wish to discontinue, hoping that the response could be maintained and, if not, could be re‐established should progression occur. The current phase 4 study was planned to evaluate the effects of drug discontinuation and retreatment with pexidartinib in patients with TGCT who previously benefited from the drug to inform clinical decision making and help guide individualized treatment strategy.

MATERIALS AND METHODS

Study design

This was an open‐label, phase 4, multicenter, global study conducted at 15 sites across the United States, Europe, Australia, and Asia (ClinicalTrials.gov identifier NCT04526704; EudraCT Number, 2020‐000192‐20). It focused on evaluating pexidartinib in patients diagnosed with TGCT who were already undergoing treatment and experiencing positive outcomes in one of the four previous studies exploring the effect of pexidartinib on TGCT. These patients were enrolled after the conclusion of those studies. The study protocol was approved by the institutional review board or independent ethics committees at each study center in accordance with the Declaration of Helsinki and Good Clinical Practice guidelines. The planned study duration was 2 years. The data from this study are from the last data cutoff, August 17, 2023.

Key eligibility criteria

Patients included in the study were aged 18 years or older with a pathologically confirmed diagnosis of TGCT. Patients were currently enrolled and benefiting from pexidartinib treatment in one of the following studies: PLX108‐10 (ENLIVEN), PLX108‐01 (phase 1 solid tumors; ClinicalTrials.gov identifier NCT01004861), PL3397‐A‐A103 (phase 1 Asian patients; ClinicalTrials.gov identifier NCT02734433), or PL3397‐A‐U126 (phase 1 open‐label pharmacokinetics study; ClinicalTrials.gov identifier NCT03291288). Patients enrolled in this study were receiving treatment at the time of screening and had disease control. Patients were screened at the end‐of‐treatment visit of their previous trial, at which time they could enroll in this study in either the treatment continuation cohort (i.e., continue treatment) or the treatment‐free/retreatment cohort (i.e., discontinue pexidartinib with the option of retreatment). All patients signed an institutional review board/institutional ethical committee‐approved informed consent form.

Treatment

Patients who chose to continue oral pexidartinib in the treatment continuation cohort continued at the same dose they were receiving in the previous trial (400 mg, 600 mg, or 800 mg daily). Patients could discontinue pexidartinib for any reason. Patients who chose to discontinue pexidartinib stopped dosing at entry into the study (i.e., after their final dose at the previous study’s end‐of‐treatment visit) and entered the treatment‐free/retreatment cohort. Patients in the treatment‐free/retreatment cohort remained treatment‐free until reinitiation of pexidartinib based on the investigator’s and/or patient’s decision, which could include reasons such as tumor progression or subjective and/or functional measures. For patients who restarted pexidartinib in the retreatment period of the treatment‐free/retreatment cohort, pexidartinib was restarted at the dosage level at which the patient completed the prior trial. For both cohorts, pexidartinib (200‐mg capsule) was dosed on an empty stomach (i.e., at least 1 hour before or 2 hours after a meal or snack). During retreatment, investigators ensured weekly liver monitoring tests for the first 8 weeks, then every 2 weeks for the next month, then every 3 months or more frequently as directed by the investigator.

Assessments

Radiologic tumor assessments, patient‐reported outcomes (PROs), and safety were assessed every 3 months. The treatment‐free period of the treatment‐free/retreatment cohort had an additional assessment at 1 month. Investigators evaluated magnetic resonance imaging scans to perform tumor assessment according to RECIST v1.1 for patients in the treatment‐continuation and treatment‐free/retreatment cohorts. Specifically, data regarding evidence of progressive disease (PD) according to RECIST v1.1 or not (no PD) were collected. There was no provision for collecting data on the presence of complete response, partial response, or stable disease. Additional assessments could be conducted if necessary. All PROs were assessed using the Patient‐Reported Outcomes Measurement Information System–Physical Function (PROMIS‐PF) and EuroQol 5‐dimension, 5‐level (EQ‐5D‐5L) questionnaires. Ongoing safety included serum chemistry, hematology, urinalysis, coagulation, electrocardiogram, hormone testing (follicle‐stimulating hormone and thyroid‐stimulating hormone), and physical examination. Patients were monitored for AEs and treatment‐emergent AEs (TEAEs). Events were coded using the Medical Dictionary for Regulatory Activities, version 26.0 and were graded using the National Cancer Institute’s Common Terminology Criteria for Adverse Events.

The primary objective was the proportion of patients who were treatment‐free at months 12 and 24 in the treatment‐free/retreatment cohort. Secondary objectives were the mean change from baseline in PROMIS‐PF and EQ‐5D‐5L visual analog scale (VAS) scores, the incidence of AEs and serious AEs, electrocardiograms, and laboratory assessments in the treatment‐continuation and treatment‐free/retreatment cohorts. Secondary objectives for the treatment‐continuation and retreatment period of the treatment‐free/retreatment cohort only were tumor progression by RECIST v1.1 according to investigator assessment and TEAEs. Progression‐free survival was defined as the time from informed consent to the earliest date of PD (according to RECIST v1.1) or death from any cause. In the absence of an event at the analysis cutoff date, patients were censored at the last adequate tumor evaluation date before the cutoff.

Statistics

This was a descriptive, hypothesis‐generating study; as such, no formal comparisons were made. The Kaplan–Meier product‐limit method was used to estimate the proportion of patients who remained treatment‐free at 12 and 24 months, as well as progression‐free survival (by cohort); medians and quartiles were based on the Kaplan–Meier method. All 95% confidence intervals (CIs) were calculated with the Brookmeyer and Crowley method. The Kaplan–Meier estimates, including the number of patients at risk and with an event, and the proportion of treatment‐free patients along with the 95% CIs, were also summarized at 3, 6, 9, 12, 15, 18, 21, and 24 months. Estimates of the proportion were provided with corresponding 95% CIs using the Greenwood formula.

RESULTS

Patient disposition is illustrated in Figure 1 and Figure S1. Between October 2020 and April 2021, in total, 50 patients were still participating in one of the four previous pexidartinib studies. Of these 50 patients, 18 patients did not enroll in this study because they switched to commercial pexidartinib (n = 10), previously discontinued pexidartinib (n = 6), and decided not to participate (n = 2). Upon completion/discontinuation of their prior studies, the sponsor no longer monitored outcomes of these patients. Of the total 32 patients enrolled in the study, there were 21 patients in the treatment‐continuation cohort and 11 in the treatment‐free/retreatment cohort. Four patients (12.5%) discontinued the study early because of withdrawal by the patient (n = 2), AE (n = 1), or physician decision (n = 1).

FIGURE 1.

Study schema and patient flow. Patients were enrolled from one of four previous studies. Patients who reinitiated pexidartinib underwent weekly liver monitoring tests for the first 8 weeks, then every 2 weeks for 1 month, then every 3 months or as directed by the investigator. ^aAt screening/baseline, patients were given their choice of cohort. Patients reinitiated on pexidartinib underwent weekly liver monitoring tests for the first 8 weeks, then every 2 weeks for 1 month, then every 3 months or as directed by the investigator. D/C, discontinued; EQ‐5D‐5L, EuroQol 5‐dimension, 5‐level visual analog scale; PD, progressive disease; PROMIS‐PF, Patient‐Reported Outcomes Measurement Information System–Physical Function; PRO, patient‐reported outcome; Q3M, every 3 months.

Demographics and baseline characteristics are shown in Table 1. Overall, the median patient age was 47.5 years, 84.4% of patients were White, and 3.1% were Hispanic or Latino; these characteristics were relatively balanced between the cohorts. Fifty percent of patients were women, and the proportion of women was higher in the treatment‐continuation cohort (57.1%) compared with the treatment‐free/retreatment cohort (36.4%). The median duration of pexidartinib treatment was 55.7 months before enrollment. Most patients (28 of 32; 87.5%) had achieved an objective response in the prior trial. Overall, the most frequent total daily dose last prescribed was 800 mg (15 of 32 patients; 46.9%) and was similar between the cohorts. In the treatment‐continuation cohort, 10 patients (47.6%) continued pexidartinib at a daily dose of 800 mg, six (28.6%) continued at 600 mg, and five (23.8%) continued at 400 mg. The average time from diagnosis to study entry was 10.6 years and was longer for patients in the treatment‐continuation cohort (10.8 years) than the treatment‐free/discontinuation cohort (6.0 years). Most tumors were in the lower joints (84.4%), with the knee being the most common (50%), and this was similar between the cohorts. A higher proportion of patients in the treatment‐free/discontinuation cohort had diffuse disease (72.7%) compared with the treatment‐continuation cohort (47.6%). The most common concomitant medications at time of study entry were agents acting on the renin‐angiotensin system (28.1%), analgesics (25.0%), and anti‐inflammatory/antirheumatic products (25%; see Table S1).

TABLE 1.

Demographic and baseline characteristics.

	No. (%)
	Treatment continuation cohort, n = 21	Treatment‐free/retreatment cohort, n = 11	Total, N = 32
Country
Australia	1 (4.8)	2 (18.2)	3 (9.4)
Hungary	2 (9.5)	0 (0.0)	2 (6.3)
Italy	4 (19.0)	3 (27.3)	7 (21.9)
Netherlands	1 (4.8)	0 (0.0)	1 (3.1)
Spain	4 (19.0)	2 (18.2)	6 (18.8)
Taiwan	1 (4.8)	1 (9.1)	2 (6.3)
United States	8 (38.1)	3 (27.3)	11 (34.4)
Age, years
Mean ± SD	46.2 ± 15.67	51.9 ± 15.21	48.2 ± 15.51
Median [range]	47.0 [21–78]	51.0 [27–81]	47.5 [21–81]
Sex
Male	9 (42.9)	7 (63.6)	16 (50.0)
Female	12 (57.1)	4 (36.4)	16 (50.0)
Ethnicity
Hispanic or Latino	1 (4.8)	0	1 (3.1)
Not Hispanic or Latino	19 (90.5)	10 (90.9)	29 (90.6)
Not collected per local regulations	1 (4.8)	1 (9.1)	2 (6.3)
Race
Asian	1 (4.8)	1 (9.1)	2 (6.3)
White	19 (90.5)	8 (72.7)	27 (84.4)
Not collected per local regulations	1 (4.8)	2 (18.2)	3 (9.4)
Weight, kg
Mean ± SD	91.52 ± 30.868	95.77 ± 26.235	92.98 ± 28.999
Median [range]	88.80 [43.0–176.3]	95.00 [44.0–150.0]	89.15 [43.0–176.3]
Median [range] duration of pexidartinib treatment in prior trial, months	55.3 [26.7–91.0]	56.7 [52.3–74.1]	55.7 [26.7–91.0]
Best response before entering the study
Complete response	6 (28.6)	5 (45.5)	11 (34.4)
Partial response	14 (66.7)	3 (27.3)	17 (53.1)
Stable disease	1 (4.8)	3 (27.3)	4 (12.5)
Total daily dose last prescribed on prior trial
800 mg	10 (47.6)	5 (45.5)	15 (46.9)
600 mg	6 (28.6)	2 (18.2)	8 (25.0)
400 mg	5 (23.8)	4 (36.4)	9 (28.1)
Time from diagnosis to informed consent, years
Mean ± SD	11.9 ± 7.5	8.1 ± 4.2	10.6 ± 6.7
Median [range]	10.8 [4.61–32.6]	6.0 [4.8–15.9]	7.8 [4.6–32.6]
Tumor subtype
Diffused	10 (47.6)	8 (72.7)	18 (56.3)
Localized	11 (52.4)	3 (27.3)	14 (43.8)
Tumor joint location
Lower	18 (85.7)	9 (81.8)	27 (84.4)
Knee	10 (47.6)	6 (54.5)	16 (50.0)
Ankle	4 (19.0)	2 (18.2)	6 (18.8)
Hip	3 (14.3)	0 (0.0)	3 (9.4)
Foot	1 (4.8)	1 (9.1)	2 (6.3)
Upper	3 (14.3)	2 (18.2)	5 (15.6)
Shoulder	1 (4.8)	1 (9.1)	2 (6.3)
Hand	1 (4.8)	0 (0.0)	1 (3.1)
Spine	1 (4.8)	0 (0.0)	1 (3.1)
Wrist	0 (0.0)	1 (9.1)	1 (3.1)
Affected joint location
Right	12 (57.1)	6 (54.5)	18 (56.3)
Left	8 (38.1)	5 (45.5)	13 (40.6)
Not applicable	1 (4.8)	0 (0.0)	1 (3.1)

Abbreviation: SD, standard deviation.

In the treatment‐free/retreatment cohort, during the treatment‐free period, six of the 11 patients (54.5%) had PD according to RECIST v1.1. The median progression‐free survival of the treatment‐free period was 22.8 months (95% CI, 1.6 months to not estimable; Figure 2A). Most (five of six) instances of PD that occurred did so within the first 12 months of discontinuation. Three of the 11 patients (27.3%) in the treatment‐free/retreatment cohort restarted pexidartinib; the probability of remaining treatment‐free at 12 and 24 months (i.e., the primary end point of the study) was 73% (95% CI, 37%–90%; Figure 2B). The primary reasons for the three patients to restart pexidartinib were symptomatic progression without RECIST v1.1 PD (n = 1), PD according to RECIST v1.1 (n = 1), and PD according to RECIST v1.1 with symptomatic progression (n = 1). All patients who restarted pexidartinib (n = 1 at each dose) achieved new disease stabilization within 6 months of retreatment (Table 2). In the treatment continuation cohort, no PD was observed for patients over the 24‐month study period (Table S2).

FIGURE 2.

Kaplan–Meier curves of (A) median progression‐free survival and (B) treatment‐free time in the treatment‐free/retreatment cohort. An event (i.e., not remaining treatment‐free) was defined as either resuming pexidartinib treatment, death (any cause), or taking systemic therapy or undergoing surgery for the treatment of TGCT, whichever occurred first. Patients who withdrew consent or were lost to follow‐up without experiencing an event were censored at the last contact date. CI, confidence interval; TGCT, tenosynovial giant cell tumor.

TABLE 2.

Overall tumor response in the treatment‐free/retreatment cohort—retreatment period, n = 3.

Overall response	Retreatment cohort, No. (%)
	Month 3 of retreatment	Month 6 of retreatment	Month 9 of retreatment	Month 12 of retreatment	Month 15 of retreatment	Month 18 of retreatment	Any progressive disease a
No.	3	3	3	3	3	1	1 (33.3)
Not progressive	2 (66.7)	3 (100.0)	3 (100.0)	3 (100.0)	3 (100.0)	3 (100.0)
Progressive disease	1 (33.3)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
Not evaluable	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)

Note: Percentages are based on the number of patients (n/N) in the analysis set within each cohort. Percentages are also based on the number of patients in the analysis set with data available (no.) within each cohort and period, if applicable, unless stated otherwise. Baseline in the retreatment period of the treatment‐free/retreatment cohort is defined as the last assessment before or on the date of the first dose of pexidartinib when restarting treatment. End of treatment refers to the assessment once pexidartinib administration is completed (end of study) or discontinued (early termination) for treated patients.

^a Patients who have had at least one postbaseline magnetic resonance image with progressive disease.

Treatment exposure and compliance are shown in Table S3. The median compliance rates were 98.8% in the treatment continuation cohort over a median duration of 22.9 months and 99.5% in the treatment‐free/retreatment cohort over a median duration of 16.2 months.

Average scores over time for PROs are shown in Figure 3A,D. Overall, mean PROMIS‐PF and EQ‐5D‐5L VAS scores remained stable over time for both cohorts. For the four patients who had RECIST v1.1 PD but did not reinitiate treatment, one patient had a 10‐point decrease in the PROMIS‐PF score at the month‐24 visit; no patients had clinically significant decreases (≥10 points) from baseline on the EQ‐5D‐5L VAS (Figure 3B,E). Of the three patients who reinitiated treatment, two had clinically significant (≥10‐point) improvements in PROMIS‐PF and EQ‐5D‐5L VAS scores after retreatment (Figure 3C,F).

FIGURE 3.

Patient‐reported outcomes. Mean PROMIS‐PF scores for (A) the treatment‐continuation cohort and the treatment‐free/retreatment cohort during the treatment‐free period, (B) patients in the treatment‐free period who had PD by RECIST v1.1, and (C) retreated patients. Mean EQ‐5D‐5L VAS scores for (D) the treatment continuation cohort and the treatment‐free/retreatment cohort during the treatment‐free period, (E) patients in the treatment‐free period who had PD by RECIST v1.1, and (F) retreated patients. Higher PROMIS‐PF or EQ‐5D‐5L VAS scores indicate a better health state. Baseline in the treatment continuation cohort was defined as the last assessment before or on the date (irrespective of time) of the first dose of pexidartinib in this study. Baseline in the treatment‐free period of the treatment‐free/retreatment cohort was defined as the value before or on the screening/baseline visit. BL, baseline; CI, confidence interval; EQ‐5D‐5L VAS, EuroQol 5‐dimension, 5‐level visual analog scale; M, month; PROMIS‐PF, Patient‐Reported Outcomes Measurement Information System–Physical Function; RECIST, Response Evaluation Criteria in Solid Tumors.

A summary of TEAEs in the treatment‐continuation and treatment‐free/retreatment cohorts is provided in Table 3, and a list of TEAEs by type and severity is shown in Table 4. Grade ≥3 AEs were reported in seven patients (33.3%) in the treatment‐continuation cohort and in one patient (33.3%) in the treatment‐free/retreatment cohort. There were no deaths during the study. There were two serious TEAEs in the treatment‐continuation cohort (Creutzfeld–Jakob disease and dislocation of a hip prosthesis) that were not related to treatment. Two patients in the treatment‐continuation cohort had TEAEs leading to discontinuation: the patient with Creutzfeld–Jakob disease and another patient with lymph gland infection. Creutzfeld–Jakob disease was not resolved at last contact. The event of lymph gland infection was grade 1 and was considered related to pexidartinib. At last contact, the AE was not resolved.

TABLE 3.

Incidence of treatment‐emergent adverse events.

	No. (%)
	Treatment continuation cohort, n = 21	Treatment‐free/retreatment cohort–retreatment period, n = 3	All pexidartinib treated, N = 24
Any TEAE	19 (90.5)	3 (100.0)	22 (91.7)
Any TEAE by worst CTCAE grade:
Grade 5, death	0 (0.0)	0 (0.0)	0 (0.0)
Grade 4, life‐threatening	1 (4.8)	1 (33.3)	2 (8.3)
Grade 3, severe	6 (28.6)	0 (0.0)	6 (25.0)
Grade 2, moderate	12 (57.1)	2 (66.7)	14 (58.3)
Grade 1, mild	0 (0.0)	0 (0.0)	0 (0.0)
Any TEAE related to study treatment	14 (66.7)	3 (100.0)	17 (70.8)
Any serious TEAE	2 (9.5)	0 (0.0)	2 (8.3)
Any serious TEAE related to study treatment	0 (0.0)	0 (0.0)	0 (0.0)
Any TEAE leading to:
Study drug discontinuation/drug withdrawn	2 (9.5)	0 (0.0)	2 (8.3)
Drug interrupted	7 (33.3)	1 (33.3)	8 (33.3)
Dose reduction	0 (0.0)	1 (33.3)	1 (4.2)
Any TEAE of special interest	0 (0.0)	0 (0.0)	0 (0.0)
Any TEAE leading to death	0 (0.0)	0 (0.0)	0 (0.0)
Any TEAE associated with COVID‐19	8 (38.1)	3 (100.0)	11 (45.8)
Any TEAE by time period:
0 to ≤6 months	15 (71.4)	3 (100.0)	18 (75.0)
>6 to ≤12 months	15 (71.4)	3 (100.0)	18 (75.0)
>12 to ≤18 months	15 (71.4)	2 (66.7)	17 (70.8)
>18 to 24 months/end of study	8 (38.1)	0 (0.0)	8 (33.3)
First TEAE by time period:
0 to ≤6 months	15 (71.4)	3 (100.0)	18 (75.0)
>6 to ≤12 months	2 (9.5)	0 (0.0)	2 (8.3)
>12 to ≤18 months	1 (4.8)	0 (0.0)	1 (4.2)
>18 to 24 months/end of study	1 (4.8)	0 (0.0)	1 (4.2)

Note: Percentages are based on the number of patients (n/N) in the analysis set within each cohort in the respective period, if applicable. A TEAE is a new adverse event or pre‐existing condition that worsens in CTCAE grade after the first dose of study drug received in this study and up to 30 days after last dose of study drug. An AESI is an event meeting the hepatic reporting criteria per protocol. AEs are considered associated with COVID‐19 when their AE number is reported in any instance of the COVID‐19 eCRF. Time period for an AE is defined based on the number of months between the first date of study treatment in this study and the AE start date.

Abbreviations: AE, adverse event; AESI, adverse event of special interest; CTCAE, Common Terminology Criteria for Adverse Events; eCRF, electronic case report form; TEAE, treatment‐emergent adverse event.

TABLE 4.

Treatment‐emergent adverse events by grade and frequency occurring in ≥10% of patients.

Preferred term	No. (%)
	Worst CTCAE grade	Treatment continuation cohort, n = 21	Treatment‐free/retreatment cohort, n = 3	All pexidartinib treated, N = 24
Blood creatine phosphokinase increased	Any grade	8 (38.1)	2 (66.7)	10 (41.7)
	Grade 4	1 (4.8)	1 (33.3)	2 (8.3)
	Grade 3	2 (9.5)	0 (0.0)	2 (8.3)
	Grade 2	2 (9.5)	0 (0.0)	2 (8.3)
	Grade 1	3 (14.3)	1 (33.3)	4 (16.7)
COVID‐19	Any grade	7 (33.3)	2 (66.7)	9 (37.5)
	Grade 2	2 (9.5)	0 (0.0)	2 (8.3)
	Grade 1	5 (23.8)	2 (66.7)	7 (29.2)
Aspartate aminotransferase increased	Any grade	5 (23.8)	1 (33.3)	6 (25.0)
	Grade 3	0	1 (33.3)	1 (4.2)
	Grade 1	5 (23.8)	0 (0.0)	5 (20.8)
Upper respiratory tract infection	Any grade	4 (19.0)	0 (0.0)	4 (16.7)
	Grade 2	4 (19.0)	0 (0.0)	4 (16.7)
Arthralgia	Any grade	4 (19.0)	0 (0.0)	4 (16.7)
	Grade 2	3 (14.3)	0 (0.0)	3 (12.5)
	Grade 1	1 (4.8)	0 (0.0)	1 (4.2)
Vomiting	Any grade	4 (19.0)	0 (0.0)	4 (16.7)
	Grade 1	4 (19.0)	0 (0.0)	4 (16.7)
Asthenia	Any grade	3 (14.3)	1 (33.3)	4 (16.7)
	Grade 2	1 (4.8)	0 (0.0)	1 (4.2)
	Grade 1	2 (9.5)	1 (33.3)	3 (12.5)
Anemia	Any grade	3 (14.3)	1 (33.3)	4 (16.7)
	Grade 2	2 (9.5)	1 (33.3)	3 (12.5)
	Grade 1	1 (4.8)	0 (0.0)	1 (4.2)
Nausea	Any grade	3 (14.3)	0 (0.0)	3 (12.5)
	Grade 2	3 (14.3)	0 (0.0)	3 (12.5)
Fatigue	Any grade	3 (14.3)	0 (0.0)	3 (12.5)
	Grade 2	1 (4.8)	0 (0.0)	1 (4.2)
	Grade 1	2 (9.5)	0 (0.0)	2 (8.3)

Note: Percentages are based on the number of patients (n/N) in the analysis set within each cohort in the respective period, if applicable. Events were coded using MedDRA version 26.0. A TEAE was a new adverse event or pre‐existing condition that worsened in CTCAE grade after the first dose of study treatment received in this study and up to 30 days after last dose of study treatment. Patients with more than 1 occurrence within a system organ class or preferred term were only counted once under each category.

Abbreviations: CTCAE, Common Terminology Criteria for Adverse Events; MedDRA, Medical Dictionary for Regulatory Activities; TEAE, treatment‐emergent adverse event.

The most common TEAEs were increased blood creatine phosphokinase (CPK; 10 patients; 41.7%), COVID‐19 (nine patients; 37.5%), and increased aspartate aminotransferase (six patients; 25.0%). A listing of treatment‐related TEAEs is provided in Table S4. Treatment‐related TEAEs were reported for 14 patients (66.7%) in the treatment‐continuation cohort and three patients (100%) during the retreatment phase in the treatment‐free/retreatment cohort. Four patients (16.7%) in the treatment‐continuation cohort reported a total of five grade ≥3 treatment‐related TEAEs, which included one patient (4.8%) with two events of a grade 3 increase in blood CPK, one patient with both a grade 4 increase in blood CPK and grade 3 pruritus, one patient (4.8%) with a grade 3 increase in alanine aminotransferase, and one patient (4.8%) with grade 3 hypertension. With respect to hepatotoxicity, no cases of Hy's law were observed (concurrent total bilirubin greater than two times the upper limit of normal and alanine or aspartate aminotransferase levels greater than or equal to three times the upper limit of normal).

Tables S5 and S6 show AEs in the treatment‐free period. The AEs occurred in seven patients (63.6%) during the treatment‐free period; all were grade ≤2.

DISCUSSION

This phase 4 study was designed to evaluate the effect of the interruption of treatment with pexidartinib in patients with TGCT who previously demonstrated clinical benefit, with the goal of informing physicians and patients about the safety and effectiveness of treatment discontinuation and retreatment. The study was designed to mimic the real‐world scenario of treatment with pexidartinib, in which patients may request to discontinue or interrupt treatment. Interestingly, most patients who entered this phase 4 study preferred to continue pexidartinib, rather than having a treatment break, despite undergoing therapy for a median of more than 4 years. In this study, 45% of patients who stopped taking pexidartinib did not show any deterioration in symptoms, nor did they exhibit radiologic progression by RECIST v1.1 at the 2‐year follow‐up after treatment cessation. PROs did not change significantly (± ≥10 points) in patients who exhibited radiologic progression. Furthermore, no instances of PD were observed in patients who continued treatment, even among those who received a reduced daily dose (i.e., 400 or 600 mg daily). All patients who decided to restart pexidartinib benefited again from the drug.

There is limited evidence on the discontinuation of CSF1R tyrosine kinase inhibition in patients with TGCT. However, such studies have shown reported progression rates after discontinuation ranging from 29% at 5 years with nilotinib, 63% after a median follow‐up of 12 months with imatinib, and 46% at 2 years with emactuzumab. ¹² , ¹³ , ¹⁴ These rates are comparable to the rate of 55% at 2 years in the current study, although differences exist with respect to response at the time of discontinuation and how disease progression was measured. In the emactuzumab study, patients were permitted to reinitiate treatment; one patient restarted treatment and was able to regain response, supporting results from our study. ¹⁴

To our knowledge, this is the first prospective study specifically designed to explore the safety and efficacy of discontinuation and reinitiation of treatment in patients with TGCT. This study, although comprising a small number of patients and including a select patient population (i.e., patients on treatment with pexidartinib who had radiologic/symptomatic improvement and acceptable toxicity), answers some important questions, among which is the question regarding the durability of the effect of pexidartinib after its discontinuation. Indeed, observing that 45% of patients who stopped taking pexidartinib did not meet RECIST v1.1 criteria for PD suggests that intermittent discontinuation could be considered as a viable option for certain patients who had previously experienced a positive tumor response. The median progression‐free survival was 22.8 months during the treatment‐free period. However, this study was not designed to provide an optimal time point at which pexidartinib interruption should occur. In this small study, the decision to interrupt pexidartinib treatment (i.e., entering the treatment‐free/retreatment cohort) was at the discretion of the physicians and patients together, based on objective and subjective measures. All patients in this study derived clinical benefit from pexidartinib treatment for TGCT and had been treated with pexidartinib for a median of 55.7 months before entering the study, when they could choose to discontinue or remain on pexidartinib. A dedicated study would be needed to determine the optimal duration of pexidartinib to achieve prolonged disease control after discontinuation. Decisions pertaining to interruption and reinitiation of pexidartinib should be shared between patients and their health care professionals.

In the 2‐year observational period, 55% of patients experienced PD after discontinuation. However, the results from the patients who were retreated suggest that disease control can be achieved again after treatment is reinitiated. Notably, only two of the six patients who had RECIST v1.1 PD restarted pexidartinib. This could have been because patients did not experience a corresponding significant symptomatic progression. Indeed, PROs remained stable in patients who showed RECIST v1.1 PD after pexidartinib discontinuation. Of the four patients who experienced PD without restarting pexidartinib, none experienced clinically significant decreases (≥10‐point) in EQ‐5D‐5L VAS scores from baseline during this period of follow‐up; one patient had a 10‐point decrease in the PROMIS‐PF score at the 24‐month visit. Two of the three patients who restarted pexidartinib did so because of symptomatic progression with corresponding decreases in EQ‐5D‐5L VAS scores. Previous studies have reported that symptoms like joint pain, swelling, and stiffness negatively affect the quality of life of patients with TGCT. ¹⁵ Of the patients who restarted pexidartinib, improvements were observed in PROMIS‐PF or EQ‐5D‐5L VAS scores that were concurrent with disease stabilization, with two patients experiencing clinically significant improvements in either PROMIS‐PF or EQ‐5D‐5L VAS scores from baseline. These results suggest that patient PROs/quality of life are important factors when considering retreatment and should be considered in combination with RECIST v1.1 progression.

The safety profile of pexidartinib in this study was consistent with results from previous studies, including ENLIVEN, ⁸ , ⁹ with no new safety signals and no cases of life‐threatening hepatotoxicity. The safety profile in the treatment‐continuation cohort demonstrates that pexidartinib was safe in patients who continued treatment across dose levels of 800 mg (n = 10; 47.6%), 600 mg (n = 6; 28.6%), and 400 mg (n = 5; 23.8%), with some receiving treatment for up to 9 years. Importantly, patients who restarted treatment after discontinuation showed tolerance to restarting treatment at their previous dose level (n = 1 each at 800, 600, and 400 mg) without hepatotoxicity. However, these are patients who had already demonstrated tolerance to pexidartinib. This study also examined CPK, kidney, and renal function parameters that were not previously studied in pexidartinib clinical studies. No instance of renal function toxicity was detected, despite up to 9 years of treatment with pexidartinib. As now reported for other CSF1R inhibitors, CPK elevation was a common event, with two of 24 (8.3%) pexidartinib‐treated patients developing a grade 4 CPK increase; however, neither event was associated with a clinically relevant effect, and both patients normalized after pexidartinib discontinuation.

Although an analysis of response by dose is beyond the scope of the study, patients who continued treatment did so at the last dose they received in the previous trial (400, 600, or 800 mg daily). Several patients in the treatment‐continuation cohort had previously received dose reductions (to 400 or 600 mg) and continued treatment at the reduced dose without disease progression. This result suggests that pexidartinib is beneficial for disease control even after dose reduction was necessary.

The strengths of this study include its design, which was intended to mimic the real‐world experience of treatment with pexidartinib; its results provide assurance to patients that pexidartinib can effectively and safely be restarted should they want or need to discontinue. These are important considerations in a chronic disease that can affect younger patients and those planning families. Limitations include a small sample size, lack of an objective definition for symptomatic progression, and lack of response precision (i.e., only recorded PD or no PD according to RECIST v1.1). This study also used a select population of patients who had previously demonstrated tolerability and efficacy at their dose of pexidartinib, so generalizability to the broader TGCT community is limited. Reasons for entering the treatment‐free/retreatment cohort were not recorded. Finally, the duration of this study was 24 months; given the nature of the design and extent of the patients’ informed consent, our interpretations are limited to this 24‐month follow‐up period. This does not exclude the possibility that other patients may experience progression after a longer interval of pexidartinib discontinuation.

CONCLUSION

The results of this study contribute meaningful information to the body of research on the safe and effective discontinuation and retreatment with pexidartinib therapy for patients with TGCT. These results should help inform physicians and patients in their treatment decisions.

AUTHOR CONTRIBUTIONS

Jayesh Desai: Conceptualization, writing–review and editing, and investigation. Andrew J. Wagner: Conceptualization, writing–review and editing, and investigation. Irene Carrasco Garcia: Writing–review and editing and investigation. Marilena Cesari: Writing–review and editing and investigation. Michael Gordon: Writing–review and editing and investigation. Chia‐Chi Lin: Writing–review and editing and investigation. Zsuzsanna Papai: Writing–review and editing and investigation. Christopher W. Ryan: Writing–review and editing and investigation. William D. Tap: Conceptualization, writing–review and editing, and investigation. Jonathan C. Trent: Writing–review and editing and investigation. Hans Gelderblom: Writing–review and editing and investigation. Peter Grimison: Writing–review and editing and investigation. Antonio López Pousa: Writing–review and editing and investigation. Brian A. Van Tine: Writing–review and editing and investigation. Maria Rubinacci: Writing–review and editing and investigation. Dong Dai: Writing–review and editing and formal analysis. Abdul Waheed Rajper: Writing–review and editing and formal analysis. Kristen Tecson: Writing–review and editing and formal analysis. Margaret Wooddell: Conceptualization, writing–review and editing, formal analysis, and data curation. Silvia Stacchiotti: Conceptualization, writing–review and editing, and investigation.

CONFLICT OF INTEREST STATEMENT

Jayesh Desai reports institutional research funding from Roche, GlaxoSmithKline, Novartis, BeiGene, Eli Lilly & Company, Bristol Myers Squibb, AstraZeneca/MedImmune, and Amgen; and personal/consulting or advisory fees from Amgen, BeiGene, Pierre Fabre, Bayer, GlaxoSmithKline, Merck KGaA, Boehringer Ingelheim, Roche/Genentech, Daiichi Sankyo Europe GmbH, Novartis, Pfizer, Ellipses Pharma, Axelia Oncology, and Amgen outside the submitted work. Andrew J. Wagner reports institutional research funding from Aadi Bioscience, Boehringer Ingelheim, Cogent Biosciences, Daiichi Sankyo, Deciphera, Eli Lilly & Company, Foghorn, and Rain Therapeutics; and personal/consulting or advisory fees from Aadi Bioscience, BioAtla, Boehringer Ingelheim, Cogent Biosciences, Daiichi Sankyo, Deciphera Pharmaceuticals, Eli Lilly & Company, InhibRx, IQVIA, Kaplan, Kymera Therapeutics, Pharma Essentia, Servier Pharmaceuticals, and Targeted Oncology outside the submitted work. Irene Carrasco Garcia reports institutional research funding for clinical studies from PharmaMar, Eli Lilly and Company, AROG, Bayer, Eisai, Lixte, Karyopharm, Deciphera, GlaxoSmithKline, Novartis, Blueprint Medicines, Nektar, Forma, Amgen, Cogent Biosciences, and Daiichi Sankyo; and personal/consulting or advisory fees and travel expenses from PharmaMar outside the submitted work. Chia‐Chi Lin reports personal/consulting or advisory fees from AbbVie, Blueprint Medicines, Boehringer Ingelheim, Bristol Myers Squibb, Daiichi Sankyo, EMD Serono Inc., F. Hoffman‐LaRoche, Merck KGaA, Novartis, and PharmaEngine; honoraria from Boehringer Ingelheim, Daiichi Sankyo, Eli Lilly & Company, Novartis, and Roche; and travel support from BeiGene, Daiichi Sankyo, Eli Lilly, and IMPACT Therapeutics outside the submitted work. Christopher W. Ryan reports institutional research funding from Ayala Pharmaceuticals, Bristol Myers Squibb, Daiichi Sankyo, Deciphera Pharmaceuticals, Exelixis, Genentech, Novartis, Karyopharm Therapeutics, Merck, Nektar, Nikang Therapeutics, Pfizer, Xynomic, Bayer, OSI, PF Argentum IP Holdings, Rain Therapeutics, Shasqi, and Synox; personal/consulting or advisory fees from Synox, Daiichi Sankyo, AVEO, Exelixis, AstraZeneca, Bristol Myers Squibb, GlaxoSmithKline, and Pfizer; and payment for expert testimony from Pfizer, GlaxoSmithKline, and Boehringer Ingelheim outside the submitted work. William D. Tap reports a standard budget for site participation in a clinical trial from Plexxikon; personal/consulting or advisory fees and travel expenses from Eli Lilly & Company and EMD Serono; personal/consulting or advisory fees from Mundipharma, C4 Therapeutics, Cogent Therapeutics, Curadey, Daiichi Sankyo, Blueprint Medicines, GlaxoSmithKline, Agios Pharmaceuticals, NanoCarrier, Deciphera Pharmaceuticals, Aadi Bioscience, Abbisko, Adcendo, Amgen, AmMax Bio, Ayala Pharmaceuticals, Ikena Oncology, IMGT, InhibRx, Ipsen Pharmaceuticals, Kowa, Pharma Essentia, Ratio Pharmaceuticals, Servier Pharmaceuticals, and Sonata; he holds a patent Companion Diagnostic for CDK4 Inhibitors (14/854,329 pending to Memorial Sloan Kettering Cancer Center/SKI); a patent Enigma and CDH18 as Companion Diagnostics for CDK4 Inhibition (SKI2016‐021‐03 pending to Memorial Sloan Kettering Cancer Center/SKI); participation on the scientific advisory board for Certis Oncology Solutions and Innova Therapeutics; stock ownership in Certis Oncology Solutions; and stock ownership as a cofounder of Atropos Therapeutics, all outside the submitted work. Jonathan C. Trent reports personal/consulting or advisory fees from Blueprint Medicines, Deciphera, Daiichi Sankyo, Epizyme, Agios Pharmaceuticals, C4 Therapeutics, Bayer, AADI Bioscience LLC, Foghorn Therapeutics, Boehringer Ingelheim, Cogent Medicine, and Servier outside the submitted work. Peter Grimison reports support for professional activities from Daiichi Sankyo Company outside the submitted work. Antonio López Pousa reports a standard budget for site participation in a clinical trial from PharmaMar, Cebiotex, GlaxoSmithKline, Eli Lilly & Company, Inmutep, and Deciphera Therapeutics outside the submitted work. Brian A. Van Tine reports grants/contracts from GlaxoSmithKline, Merck, Pfizer, Polaris Pharmaceuticals, and Tracon Pharmaceuticals; personal/consulting or advisory fees from Aadi Bioscience, Actua Capital Partners LLC, ADRx, Advenchen Laboratories, Agenus, Apexigen Inc., Bayer, Boehringer Ingelheim, Boxer Capital LLC, Curis Inc., Cytokinetics Inc., Deciphera Pharmaceuticals Inc., Daiichi Sankyo Company, EcoR1 Capital LLC, Epizyme Inc., Hinge Bio, PTC Therapeutics, Regeneron Pharmaceuticals, and Salarius Pharmaceuticals Inc.; honoraria for educational talks from Itertion Therapeutics and Total Health Conference; service on a data/safety advisory board for Apexigen, Daiichi Sankyo, Epizyme, Bayer US Medical Affairs, PTC Therapeutics, Aadi Biosciences, Boehringer Ingelheim, Agenus, Regeneron Pharmaceuticals, Advenchen, and Curtis; nonpaid service as a board member with Polaris; travel support from Adaptimmune LLC, Epizyme Inc., and Kronos Bio; and holds a patent with Accuronix Therapeutics, all outside the submitted work. Maria Rubinacci, Dong Dai, Abdul Waheed Rajper, Kristen Tecson, and Margaret Wooddell are employees of Daiichi Sankyo, Inc. Silvia Stacchiotti reports institutional support from Advenchen, Bayer, Boehringer Ingelheim, Blueprint Medicines, Daiichi Sankyo, Deciphera Pharmaceuticals, Epizyme, Eli Lilly & Company, GlaxoSmithKline, Hutchinson, Karyopharm, Novartis, PharmaMar, Rain Therapeutics, and SpringWorks Therapeutics; personal/consulting or advisory fees from Aadi, Agenus, Astex Pharmaceuticals, Bavarian Nordic A/A, Bayer, Boehringer Ingelheim, Daiichi Sankyo, Deciphera Pharmaceuticals, Epizyme, Gentili, GlaxoSmithKline, Ibsen Innovation, Ikena, Maxivax, NEC Oncoimmunity, Novartis, PharmaMar, Rain Therapeutics, Regeneron Pharmaceuticals, and Servier Pharmaceuticals; and support for other business activities from Aadi and SpringWorks outside the submitted work. The remaining authors disclosed no conflicts of interest.

Supporting information

Supplementary Material

Desai J, Wagner AJ, Carrasco Garcia I, et al. A phase 4, multicenter, global clinical study to evaluate discontinuation and rechallenge of pexidartinib in patients with tenosynovial giant cell tumor previously treated with pexidartinib. Cancer. 2025;e35634. doi: 10.1002/cncr.35634

DATA AVAILABILITY STATEMENT

Anonymized, individual participant data on completed studies and applicable supporting clinical study documents may be available upon request at https://vivli.org. For cases in which clinical study data and supporting documents are provided pursuant to our company policies and procedures, Daiichi Sankyo will continue to protect the privacy of the company and the clinical study participants. Details on data sharing criteria and the procedure for requesting access can be found at this web address: https://vivli.org/ourmember/daiichi‐sankyo.