NTRK fusions in Lung Cancer: From Biology to Therapy

Abstract

Fusions involving the TRK protein tyrosine kinases are oncogenic drivers in a variety of tumors in children and adults, with a prevalence of ~0.2% in non-small cell lung cancer. Diagnosis can be challenging due to structural features such as NTRK intron length, but next-generation sequencing (NGS), including RNA-based NGS, increases detection. The first-generation TRK inhibitors, larotrectinib and entrectinib, have demonstrated clinically meaningful antitumor activity in TRK fusion-positive cancers in a tumor-agnostic fashion and should be considered first-line therapeutic options for TRK fusion-positive lung cancers. Furthermore, the first-generation TRK inhibitors are well tolerated. Care should be taken, however, to monitor on-target adverse events, such as dizziness, weight gain, paresthesias, and withdrawal pain. On-target and off-target mechanisms mediating TRK inhibitor resistance may occur. Next-generation TRK inhibitors, such as selitrectinib, repotrectinib, and taletrectinib, are available on ongoing clinical trials and have been shown to address on-target resistance. This review will focus on NTRK fusions and TRK-directed targeted therapy specifically in the context of lung cancer.

INTRODUCTION

As recently as a decade and a half ago, patients diagnosed with non-small cell lung cancer (NSCLC) had dismal survival outcomes. Breakthroughs in cancer biology have since resulted in the identification of an impressive number of actionable molecular alterations that match patients with NSCLC to highly active targeted therapies. Outstanding survival rates are now attainable for patients with oncogenic kinase driven NSCLCs.

There are three main genomic mechanisms that lead to abnormal tyrosine kinase activation in lung cancer – mutations, copy number changes, and fusions. While the activity of targeted therapy in NRG1, BRAF, and ERBB family fusions is under exploration, activating fusions involving ALK, ROS1, and RET match patients to tyrosine kinase inhibitors that are approved by one more health care agencies [1]. NTRK fusions joined the latter group in 2018.

BIOLOGY

Wildtype TRK.

There are three TRK tyrosine kinase receptors, TRKA, TRKB, and TRKC. These are encoded by the genes NTRK1, NTRK2, and NTRK3, respectively. The TRK pathway has an important role in embryonic neuronal development and differentiation. In adults TRK expression can be encountered in neural and muscular tissue. Neurotrophin ligand binding and TRK activation and result in homodimerization of the receptor, followed by transactivation of the intracellular domains, and recruitment of cytoplasmic adaptors. These adaptors, in turn, activate downstream signaling via the MAPK, PI3K, and/or PKC pathways [2].

TRK in oncogenesis.

Actionable oncogenic TRK activation is primarily mediated by NTRK gene fusion; NTRK mutation and amplification are not known to be highly actionable in the clinic. NTRK fusions are formed by intrachromosomal or interchromosomal rearrangements in which 3′ sequences of NTRK1, NTRK2, or NTRK3 are colligated with 5′ sequences of other genes. The resulting chimeric TRK kinase containing oncoproteins are characterized by ligand independent constitutive activation by a variety of mechanisms, including aberrant dimerization and increased TRK kinase expression [2].

CLINICOPATHOLOGIC FEATURES

TRK fusions across all cancers.

NTRK fusions are found across a variety of adult and pediatric cancers [2]. In terms of frequency, these fusions can be (1) highly prevalent in rare cancers or (2) much less prevalent in more common cancers. Cancers in the former group include secretory cancers of the salivary gland (formerly known as mammary analogue secretory carcinomas) or breast, congenital fibrosarcoma, and subtypes of congenital mesoblastic nephroma. NTRK fusions can be found in these cancers in >90% of cancers, making them almost pathognomonic of these disease states. Cancers in the latter group include sarcomas, thyroid cancers, gastrointestinal cancers, non-secretory breast cancers, and lung cancers. NTRK fusions are found at varying frequencies (most below 1%) [3].

TRK fusions in lung cancer.

The prevalence of NTRK fusions in NSCLC is estimated at 0.1% to 0.3% among patients with NSCLC [4], far less common than other oncogenic fusions involving ALK (5-7%), RET(1-2%) and ROS1 (1-2%) [5–7]. A meta-analysis of NTRK fusions in various tumor histologies estimated the frequency in NSCLC of 0.17% (95% confidence interval [CI], 0.09–0.25) [4], comparable to a separate study reporting a 0.23% frequency among 4,872 patients with NSCLC. TRK fusions are enriched in the absence of canonical driver mutations [3]. NTRK1 fusions, for example, were identified in 3.3% of patients with NSCLC without known oncogenic alterations [8].

Data on the frequency of NTRK fusions across the three genes are sparse. In one small series of eleven patients, seven had NTRK1 fusions, and four had NTRK3 fusions. Fusions involving NTRK2 fusions may be rarely found in lung cancer; these fusions appear to be more enriched in central nervous system tumors [2]. Whereas most patients have adenocarcinoma, NTRK fusions are also identified in squamous cell carcinoma (SCC) and neuroendocrine carcinoma [9]. NTRK fusions tend to occur de novo in a mutually exclusive fashion with other canonical lung cancer drivers. In other oncogene-drive lung cancers, however, these fusions can emerge as resistance mechanisms to tyrosine kinase inhibitor (TKI) therapy. For example, the emergence of NTRK fusions in response to EGFR TKI therapy has been described in EGFR-mutant NSCLCs [10].

While most patients with NTRK fusion-positive lung cancers share similar clinical features (a younger median age and a minimal or no prior history of cigarette smoking) with ALK, ROS1, or RET fusion-positive lung cancers, NTRK fusions are identified in diverse patients of varying ages and prior smoking histories. These data highlight the importance of molecular profiling regardless of clinical features.

DIAGNOSIS

DNA-based multi-gene sequencing.

Broad molecular profiling of NSCLCs at diagnosis is paramount; the NCCN Guidelines recommend that eight gene groups including NTRK that are associated with actionable alterations be interrogated. Next-generation sequencing (NGS) is less costly than performing single-gene testing and identifies the largest number of patients who are candidates for targeted therapy [11]. Furthermore, the co-mutational landscape and other markers such as tumor mutational burden can be elucidated.

Current NGS technologies employ different sequencing methods. Library preparation is an important consideration for fusion detection: hybrid-capture and amplicon-based methods offering distinct advantages and limitations. Hybrid-capture library preparation is preferred as it allows the detection of known and novel fusions. While nucleic acid–input requirements may be lower, amplicon-based library preparation only detects known fusions.

RNA-based multi-gene sequencing.

DNA sequencing assays provide no direct evidence that rearrangements produce a fusion expressed at the mRNA level. This can be important as functional fusion transcripts may or may not be generated, potentially contributing to variable response to targeted therapy [12]. Detection of NTRK2 and NTRK3 fusions can also be challenging, as the common fusion breakpoints reside within large intronic regions, making capture and sequencing technically infeasible. DNA-based NGS may also find equivocal fusion events that require clarification. These equivocal calls include noncanonical fusions harboring rare or not previously annotated partners or breakpoints, or reciprocal fusions [13].

To circumvent these caveats, RNA-based NGS maximizes the likelihood of identifying NTRK fusions. This enables distinction of in-frame, transcribed gene fusions and avoids the challenges of sequencing large intronic regions. RNA sequencing can be done via targeted or less commonly whole transcriptome sequencing. Anchored multiplex PCR, an amplicon-based targeted RNA sequencing approach, allows the detection of gene fusion transcripts without prior knowledge of 5′ fusion partners and breakpoints, using nested, unidirectional gene-specific primers on one side, and universal primers on the other side. In one lung adenocarcinoma patient series for which no known driver was previously identified on prior DNA-based NGS, RNA sequencing enabled the identification of a variety of gene fusions, including NTRK2 and NTRK3, which were found in 5% and 2.7% of patients, respectively [14].

Single- or oligo-gene testing.

Limited or single gene testing has become less favored as a primary approach, but can occasionally be leveraged in situations where a high pretest probability of finding an NTRK fusions is present (i.e. secretory carcinoma of the salivary gland) and/or tissue is scant. Fluorescence in situ hybridization (FISH) is a DNA-based technique which has historically been used for specific fusion detection. It requires three sets of break-apart probes (one for each NTRK gene) or uses several specific probes for known NTRK fusion partners (e.g. ETV6). Reverse transcriptase (RT)-PCR is another method that could be employed for the detection of specific fusion events; it is seldomly used given the wide variety of NTRK fusions.

Protein expression analysis.

Immunohistochemistry (IHC) is a rapid and inexpensive diagnostic screening tool. The most used antibody (rabbit recombinant monoclonal antibody, clone EPR17341, Abcam, Cambridge, MA) detects all three TRK proteins (TRKA, TRKB and TRKC) and recognizes a C-terminal epitope in the tyrosine kinase domain. IHC is not specific for the TRK fusion protein and also detects wild-type TRK expression. Different subcellular staining patterns can be considered positive, such as cytoplasmic, membranous, nuclear, and perinuclear, varying by the type of upstream gene partner. The most commonly used criteria for positivity is staining of at least 1% of tumor cells at any intensity above background [15]. However, other criteria are also used and a consensus has not yet been reached on how best to define TRK fusion positivity by IHC [16]. Despite the good sensitivity and specificity of IHC-based detection methods, orthogonal confirmatory nucleic acid-based testing should be performed when possible [17].

Circulating tumor DNA testing.

Obtaining an adequate tissue sample can be challenging and can result in inadequate tumor cellularity for molecular analysis. Sequencing of circulating tumor DNA can be used to complement tumor sequencing and provides the added benefit of accounting for intra- and inter-tumor heterogeneity. The sensitivity of detecting NTRK fusions in plasma is lower than in tissue; however, new algorithms for identifying fusions are in development to increase detection rates of NTRK1 fusions in plasma [18]. Plasma testing may be of substantial use in the setting of TKI resistance as resistance mutations can be detected more easily.

Overall testing strategy.

In summary, an approach combining hybrid capture DNA-based NGS and RNA-based NGS is the preferred method for detecting NTRK fusions; the former elucidates the co-mutational landscape broadly and the latter improves diagnostic sensitivity. In practice environments where such testing is not covered or cannot be accessed, IHC could be considered with orthogonal validation when possible.

TRK INHIBITOR ACTIVITY

Larotrectinib.

Larotrectinib is a first-in-class, orally bioavailable, and highly selective inhibitor of all three TRK protein kinases. It is a small molecule designed to potently block the ATP binding site of all TRK proteins [half maximal inhibitory concentration (IC₅₀) values of 5–11nM] [19, 20]. In adults, the recommended dose of larotrectinib is 100 mg orally twice daily, with or without food, and it is available in a capsule form and as an oral solution (20 mg/mL).

The antitumor activity of larotrectinib has remained remarkably consistent over time and with increased patient accrual. In the first 55 patients consecutively enrolled in three phase I/II clinical trials (NCT02122913, NCT02637687, and NCT02576431) of TRK fusion-positive cancers was initially published in 2018. The overall response rate (ORR) was 80% (95% CI, 67 to 90) according to investigator assessment [21]. After extended follow-up (median of 32.5 months), larotrectinib demonstrated a median duration of response (DOR) of 35.2 months (95% CI 19.8-not estimable [NE]) and a median progression-free survival (PFS) of 25.8 months (95% CI 9.9-NE). The median overall survival (OS) was not reached (95% CI 44.4-NE) [22]. With an additional 163 patients, the ORR from 218 patients with tumors harboring NTRK fusion treated with larotrectinib was 75% (95% CI 68% to 81%]), with complete responses in 22% [23]. Further exploratory analysis showed good concordance of investigator assessment with blinded independent central review suggesting larotrectinib has a durable response profile.

A separate analysis of patients with lung cancer was performed (Table 1). A total of 20 patients with NTRK fusion-positive lung cancers were identified. The median age was 48 years (range 25-76), with equal distribution between men (50%) and women (50%). NTRK1 fusion was found in 16 patients (80%), and four patients had NTRK3 fusions (20%). The ORR was 73% (95% CI 45-92). The median DOR was 33.9 months (95% CI 5.6 – 33.9). The median PFS and OS were 35.4 months (95% CI 5.3- 35.4) and 40.7 months (95% CI 17.2 - NE), respectively (Table 2) [24].

TABLE 1. Clinicopathologic features of NTRK fusion-positive lung cancers treated on registrational trials.

The clinicopathologic features of patients with advanced NTRK fusion-positive lung cancers who were treated in the regulatory data sets of the larotrectinib and entrectinib programs are summarized. These features are comparable to previously published data in descriptive cohorts of NTRK fusion-positive lung cancers.

	Larotrectinib (n=20) [24]	Entrectinib (n=13) [27, 32]
Age, median (range)	48.5 (25-76) years	60 (46-77) years
CNS metastases at baseline, n (%)
No	10 (50)	5 (38)
Yes	10 (50)	8* (62)
Previously treated with radiotherapy	2 (10)	5 (38)
NTRK fusion, n (%)
NTRK1	16 (80)	8 (61)
NTRK2	0	1 (8)
NTRK3	4 (20)	4 (31)
Tumor histology, n (%)
Adenocarcinoma	19 (95)	9 (69)
Squamous Cell carcinoma	0	2 (16)
Neuroendocrine carcinoma	1 (5)	0
NSCLC - NOS	0	2 (16)

NSCLC, non-small cell lung cancer; NOS, not otherwise specified

TABLE 2. Activity of first-generation TRK inhibitors in NTRK fusion-positive lung cancers.

The clinical activity of NTRK fusion-positive lung cancers who were treated in the regulatory data sets of the larotrectinib and entrectinib programs are summarized

	Larotrectinib (n=20) [24]	Entrectinib (n=13) [27]
ORR (95% CI)	73% (45–92%)	69% (39-91%)
−CR/PR rate	7%/67%	8%/61%
Median DoR, months (95% CI)	33.9 (5.6-33.9)	NE (5.6-NE)*
Median PFS, months (95% CI)	35.4 (5.3-35.4)	14.9 (4.7-NE)
Median OS, months (95% CI)	40.7 (17.2-NE)	14.9 (5.9–NE)

^*n=9

ORR, objective response rate; OS, overall survival; PFS, progression-free survival; PR, partial response; CI, confidence interval; CR, complete response; DoR, duration of response; NE, not estimable.

^*assessed by blinded independent central review

Entrectinib.

Entrectinib is an ATP-competitive inhibitor of all three TRK proteins, ROS1, and ALK, with low nanomolar concentration enzymatic efficacy (IC₅₀ values of 1-2 nM) [25]. In adults, entrectinib is orally administered at a dose of 600 mg once daily.

Efficacy data and the safety profile of entrectinib have been obtained in a pooled analysis across four phase I/II clinical trials: STARTRK-NG, ALKA-372–001, STARTRK-1, and STARTRK-2. In the integrated analysis of 74 adult patients with various NTRK fusion-positive tumors, the ORR was 63.5% (95% CI 51.5–74.4), with a median DOR of 12.9 months (95% CI 9.3–NE). The median PFS and OS were 11.2 months (95% CI 8.0–15.7) and 23.9 months (16.0–NE), respectively [26].

Among 13 patients with NTRK fusion-positive NSCLC (Table 1), eight (62%) had an NTRK1 fusions, one (7%) had an NTRK2 fusion, and four (31%) had NTRK3 fusions. Eight patients (62%) were current/former smokers and nine patients (69%) had adenocarcinoma. The ORR was 69% (95% CI 38.6-90.9), and the median DOR was not estimable (95% CI 5.6-NE). The median PFS was 14.9 months (95% CI 4.7–NE) and the median OS was 14.9 months (95% CI 5.9–NE) (Table 2) [27].

Intracranial Activity.

Clinical data demonstrate that both larotrectinib and entrectinib can cross the blood-brain barrier. Entrectinib is an effective brain penetrant molecule and, unlike larotrectinib, is a weak P-gp substrate; despite these differences, CNS activity has been observed with both drugs [28] . In a subset analysis of 14 patients with brain metastases, larotrectinib demonstrated an ORR of 71% (95% CI 42-92) [29]. In 14 evaluable patients with primary central nervous system (CNS) tumors, including children, the ORR with larotrectinib was 36%, with responses observed in various histological types of glial tumors, and the duration of treatment ranged from 0.03 months to 16.6 months [30]. From three phase I/II clinical trials of entrectinib, 16 out of 74 patients were identified with CNS metastases at baseline. The intracranial ORR in 16 patients was 50%, with four patients achieving partial response and four patients achieving complete response in the full analysis set [31].

It is noteworthy that baseline clinical characteristics of patients with lung cancer and CNS metastasis differ in the larotrectinib and entrectinib datasets. In the integrated analyses, 10 of 20 patients (50%) with lung cancer who received larotrectinib had baseline brain metastases. For entrectinib, 8 of 13 patients (62%) had CNS metastases at baseline assessed by BIRC (Table 1). The ORR of larotrectinib in patients with NSCLCs with baseline CNS metastases was 63% (95% CI 25-91), compared to 67% (95% CI 30–93) with entrectinib [24, 27, 32]. One patient with measurable intracranial disease who received larotrectinib had a 100% reduction in CNS lesions by cycle 4. The intracranial ORR in NTRK fusion-positive NSCLCs on entrectinib was 63% and in patients with measurable CNS metastases at baseline was 60%. The median intracranial PFS was 8.9 months [32] .

TRK INHIBITOR SAFETY

Overall profile.

Overall, both first-generation TRK inhibitors are well tolerated. The frequency of dose reduction with larotrectinib and entrectinib was 9% and 21%, respectively; The frequency of treatment discontinuation was <1% with larotrectinib and 4% with entrectinib [33, 34]. The most common treatment-related adverse events (AEs) were grade 1 or 2 fatigue (30%), cough (29%), and ALT increase (25%) for larotrectinib [22] and dysgeusia (41%), fatigue (25%), and constipation (23%) for entrectinib [35]. Rise in creatinine has been reported for entrectinib as a notable side effect unrelated to TRK inhibition which is thought to be secondary to entrectinib’s direct MATE1 transporter inhibition.

On-target AEs.

The TRK pathway is highly involved in the development and maintenance of the nervous system. As a result, TRK inhibitors can lead to neurologic AEs, including dizziness, weight gain, paresthesia, and TKI withdrawal pain, which occur in approximately 41%, 53%, 18%, and 34% of patients, respectively. Pharmacologic management and/or dose modification can be considered to manage these AEs as extensively reviewed elsewhere [36].

Briefly, dizziness should be characterized carefully as it can manifest as ataxia or orthostasis that can be addressed with meclizine/scopolamine and midodrine, respectively, although dose reduction was most effective in many cases that were moderate/severe. For weight gain, apart from exercise and diet, supportive medications include GLP-1 analogs. Paresthesias tend to be mild and often self-limiting. For TKI withdrawal pain mitigation, TRK inhibitor interruptions should be avoided and consider a slow tapering for patients who will discontinue therapy.

TRK INHIBITOR RESISTANCE

On-target resistance.

Despite rapid and durable responses to TRK inhibitors in lung cancer, resistance often develops. Both on-target and off-target mechanisms are observed. Paralogous to emergent resistance mutations in other oncogenic fusions (e.g. involving ALK/ROS1) after tyrosine kinase inhibition, NTRK kinase domain mutations can mediate resistance to TRK inhibitors. These mutations result in amino acid substitutions involving three major conserved regions: the solvent-front, the gatekeeper residue, and the xDFG motif. The solvent front substitutions ALK^G1202R and ROS1^G2032R are paralogous to TRKA^G595R, TRKB^G639R and TRKC^G623R; the gatekeeper substitutions ALK^L1196M and ROS1^L2026M are paralogous to TRKA^F589L, TRKB^F633L, and TRKC^F617L; and the xDFG substitutions ALKG¹²⁶⁹ are paralogous to TRKA^G667C, TRKB^G709C, and TRKC^G696A [37–39]. These alterations may induce structural changes that interfere with TRK inhibitor binding or changes that result in the preferential adoption of the inactive conformation of the kinase.

Next-generation TRK inhibitors.

Next-generation TRK inhibitors have been designed to address on-target resistance. Early phase clinical trials are underway for both selitrectinib (LOXO-195) and repotrectinib (TPX-0005). Additional agents have since been developed: taletrectinib (DS6051b/AB-106), SIM1803-1A, and PBI-200. These drugs showed in vitro and in vivo activity against NTRK wild type and mutant kinases [40–42]. Next-generation inhibitors like selitrectinib and repotrectinib feature a compact macrocyclic structure and can engage the ATP pocket without steric hindrance from TRK substitutions; this allows targeting of both wild-type and mutant kinases.

Clinical proof of principle data on the utility of next-generation TRK inhibitors after progression on prior TRK TKI therapy has been presented. Selitrectinib demonstrated an ORR of 45% (9/20) in 20 patients whose cancers harbored NTRK mutations (2 xDFG; 4 gatekeeper; 14 solvent front) [43, 44]. No responses were observed in patients with unknown mechanisms of resistance or bypass resistance. In the phase 2 portion of the ongoing TRIDENT-1 study, three out of six (50%) patients who received prior tyrosine kinase inhibitor therapy had confirmed responses on repotrectinib [45].

Off-target mechanisms.

NTRK-rearranged tumors can develop bypass pathway activation after progression on first-generation TRK inhibitors [46]. These off-target resistance mechanisms require blockade of a concurrently activated receptor tyrosine kinase and/or increased downstream signaling. Cocco et al. described different putative mechanisms including hotspot mutations or amplification involving KRAS, MET, ERBB2, or BRAF which ultimately converge to restore mitogen-activated protein kinase (MAPK) signaling. Notably, one patient, who developed MET amplification as a mechanism of off-target resistance to entrectinib was treated with selitrectinib and the multikinase MET inhibitor crizotinib, reestablishing disease control, along with the disappearance of the previously detectable MET amplification and NTRK fusion in circulating cell-free DNA (cfDNA) [46]. Additional data regarding off-target pathway activation in NSCLCs that progressed on TRK inhibition have yet to be described.

TREATMENT SEQUENCING

The treatment paradigm for metastatic NTRK fusion-positive lung cancers does not substantially differ from paradigms in other oncogene-driven contexts. If an NTRK fusion is discovered prior to the initiation of systemic therapy, larotrectinib or entrectinib should be considered as first-line treatment. If an NTRK fusion is after first-line systemic therapy was started, one can either continue therapy in the face of ongoing benefit or interrupt, or switch to larotrectinib or entrectinib should benefit be suboptimal or toxicity be a major concern.

Disease progression should be characterized as solitary site progression, oligoprogression, or widespread progression. When feasible, definitive local therapy (e.g. radiation, ablation, and/or surgery) for solitary site or oligoprogression should be considered as a means of extending the total time on larotrectinib/entrectinib. Molecular characterization can help triage patients to additional targeted therapy on a clinical trial or compassionate use program (e.g. a next-generation TRK inhibitor for selected NTRK resistance mutations or combination therapy for off-target resistance). Standard of care systemic therapy would be preferred in the absence of actionable resistance mechanisms (or complex polyclonal genomics). Thus far, the rare nature of NTRK fusion-positive lung cancers has precluded a thorough analysis of the activity of chemotherapy, chemoimmunotherapy, and immunotherapy, although the majority of these tumors may not have a high likelihood of response to immunotherapy similar to other driver-positive cancers. Continuation of TKI inhibitor therapy with chemotherapy requires exploration but has demonstrated utility in other oncogene-driven lung cancers.

Notably, the approval of larotrectinib and entrectinib allows the use of these agents in the locally advanced, non-metastatic setting.

CONCLUSIONS

NTRK-rearranged lung cancers represent a small subset of lung adenocarcinoma with similar clinicopathological features to other rearrangement-driven lung cancers. NTRK fusion testing is recommended for all advanced lung cancers at diagnosis. The inclusion of RNA-based analysis as part of comprehensive profiling that interrogates other lung cancer drivers is recommended.

TRK inhibitor therapy is a preferred first-line therapy for patients with advanced NTRK-rearranged lung cancers. First-generation TRK inhibitors achieve high response rates, CNS activity and durable disease control with favorable toxicity profiles. Unique neurologic adverse events can occur with TRK inhibition - awareness and serial monitoring is paramount.

Subsequent therapy after progression on first-generation TRK inhibitors should be individualized. Whenever feasible, repeat molecular profiling can help identify actionable resistance mechanisms. On-target resistance due to secondary mutations in the NTRK kinase domain can be overcome by next generation TRK-inhibitors that are currently in clinical trials.

HIGLIGHTS.

• NTRK fusions are bona fide oncogenic lung cancer drivers. NTRK fusion-positive lung cancers share a similar clinical profile with many other fusion-positive lung cancers.

• The approved first-generation TRK inhibitors, larotrectinib and entrectinib, achieve high response rates, intracranial coverage, and durable disease control.

• Next-generation TRK inhibitors have demonstrated clinical proof of principle responses in patients refractory to larotrectinib or entrectinib.

• The unique side effects of TRK inhibition include weight gain, dizziness, paresthesias, and kinase inhibitor withdrawal pain.