Heart rate decrease during crizotinib treatment and potential correlation to clinical response
Crizotinib is used for the treatment of advanced anaplastic lymphoma kinase (ALK)-rearranged nonsmall cell lung cancer (NSCLC). Sinus bradycardia (SB) is a side effect listed in its package insert. We investigated the frequency and timing of SB, patient characteristics associated with SB during crizotinib treatment, and potential correlation between heart rate (HR) changes and clinical response to crizotinib.
A retrospective chart review was conducted of the timing and frequency of SB, patient characteristics, and clinical response of patients to crizotinib treatment.
Forty-twp patients who had ALK-rearranged or mesenchymal epithelial transition (MET)-amplified NSCLC and received treatment with oral crizotinib 250 mg twice daily who were enrolled in 2 crizotinib trials (PROFILE 1001 and PROFILE 1005) were analyzed. There was an average decrease of 26.1 beats per minute (bpm) from the pretreatment HR among all patients during crizotinib treatment. Twenty-nine patients (69%) experienced at least 1 episode of SB (HR, <60 bpm). The average time to the lowest HR recorded was 18.6 weeks (range, 5-72 weeks). Patients who experienced SB were significantly older (aged 55.8 years vs 47.8 years; P = .0336), had a lower pretreatment HR (mean, 77.9 bpm vs 100.6 bpm; P = .002), and were on crizotinib longer (52.9 weeks vs 24.6 weeks; P = .0050) than patients who did not experience SB. The overall response rate (P = .0195) and the maximum tumor shrinkage (P = .0205) were significantly greater in patients who experienced SB.
HR decrease is common during crizotinib treatment. It remains to be determined whether the correlation between HR decrease and clinical response to crizotinib reflects a biomarker of drug efficacy or a time/cumulative dose-dependent phenomenon. Cancer 2013;119:1969–1975. © 2013 American Cancer Society.
Crizotinib is a first-in-class, orally available anaplastic kinase (ALK) inhibitor that has exhibited impressive clinical activity in ALK-rearranged nonsmall cell lung cancer (NSCLC) in 2 phase 2, single-arm clinical trials. It has also demonstrated significantly improved progression-free survival (PFS) over single-agent chemotherapy as second-line treatment for patients with ALK-rearranged NSCLC in a randomized phase 3 trial. Crizotinib has been approved in many countries for the treatment of ALK-rearranged NSCLC. In addition, as a multitargeted receptor tyrosine kinase (RTK) inhibitor that also can inhibit both the mesenchymal epithelial transition (MET) and ROS1 receptor tyrosine kinases, crizotinib has demonstrated clinical activity in MET-amplified NSCLC and ROS1-rearranged NSCLC.[5, 6] Thus, crizotinib will likely play an expanding role in targeted therapy for NSCLC.
Crizotinib is generally well tolerated, and most of its side effects are in the grade 1 and 2 range. The common side effects reported are transient visual disorders and gastrointestinal upset (nausea, vomiting, diarrhea, or constipation). It has also been reported that there was an average 2.5-beats per minute (bpm) decrease in the heart rate (HR) with every 100-ng/mL increase of serum crizotinib concentration. Indeed, according to the crizotinib package insert, grade 1 and 2 sinus bradycardia (SB) occurred in 5% of patients.
We have observed many patients who had a reduction in their HRs while receiving crizotinib treatment, and a few patients developed asymptomatic SB with an HR <45 bpm. This observation led us to perform a retrospective analysis investigating the frequency, timing, and patient characteristics associated with SB during crizotinib treatment. In addition, we performed an exploratory analysis of a potential correlation between HR changes and response to crizotinib.
MATERIALS AND METHODS
Patients with ALK-rearranged or MET-amplified NSCLC who were enrolled in the molecular expansion cohort of the PROFILE 1001 trial (National Clinical Trial [NCT] 00585195) and the PROFILE 1005 trial (NCT00932451) at the University of California Irvine Medical Center were identified and analyzed. The main eligibility criteria for both trials were aged >18 years, stage IV ALK-rearranged NSCLC, and performance status from 0 to 2. PROFILE 1001 allows any line of prior therapy, whereas PROFILE 1005 requires patients to fail at least 1 prior line of platinum-based chemotherapy. ALK rearrangement was detected by break-apart fluorescence in sit hybridization, as described previously.MET amplification was determined by FISH, and MET/chromosome enumeration probe 7 (CEP7) values >2.2 were considered amplified, as previously reported. Clinicopathologic characteristics of the patients were reviewed, including age at diagnosis, sex, disease stage, smoking history, histology, number of prior regimens, performance status, tumor response, degree of tumor shrinkage, pretreatment HR, and lowest HR achieved. Concomitant medications including, beta-blockers and calcium channel blockers, were reviewed. For patients enrolled in PROFILE 1001, disease assessment was performed every 8 weeks (2 cycles), whereas disease was assessed every 6 weeks (2 cycles) for patients enrolled in PROFILE 1005. Clinic visits were every cycle for the first 15 cycles and then every 2 cycles thereafter for both trials. All patients received crizotinib 250 mg twice daily. Both trials allowed continuation of crizotinib beyond progression according to Response Evaluation Criteria in Solid Tumors (RECIST) if there was “clinical benefit,” as determined by the investigators. Common Terminology Criteria for Adverse Event (CTCAE) version 3.0 was used to grade adverse events. The data cutoff date was February 28, 2012. All patients signed informed consent. The PROFILE 1001 and 1005 trials both were approved by the University of California Irvine Institutional Review Board.
Heart Rate Changes
CTCAE version 3 was used for both PROFILE 1001 and 1005. SB was defined as an HR <60 bpm. We defined “profound” SB as an HR <50 bpm. We chose this cutoff level, because an HR <50 bpm is generally considered by cardiologist the point at which patients potentially can experience SB symptoms, such as dizziness or syncope. All patients had pretreatment electrocardiograms (ECGs), as mandated by both clinical trials. ECGs were obtained from all patients who experienced an HR <60 bpm. ECGs were obtained at every clinic visit if the HR was ≤45 bpm for patients with profound SB even if they reported no symptoms after initial clinic visits. These ECGs were not mandated by the protocols but were obtained as part of the treatment monitoring as we observed that more patients experienced symptomatic SB. The pretreatment HR, the lowest HR achieved during treatment, the time to maximum HR decrease, vital signs (including blood pressure readings), and potential symptoms related to SB all were abstracted from clinic charts.
RECIST version 1.0 was used to measure response in PROFILE 8081001, and version 1.1 was used for PROFILE 1005. The maximum decrease in tumor measurement from the pretreatment baseline level was obtained from clinic chart. A complete response (CR), a partial response (PR), stable disease (SD), and progressive disease (PD) were defined according to RECIST.
Statistical analysis was performed using the SAS statistical software package (version 9.2; SAS Institute, Inc., Cary, NC). Chi-square analysis was used to test dichotomous variables, and the Student t test was used for continuous variables. Statistical differences were considered significant when P values were < .05.
Patient Characteristics Overall and According to Sinus Bradycardia or No Sinus Bradycardia Experienced
Forty-two patients (30 from PROFILE 1001 and 12 from PROFILE 1005) were analyzed. None of the patients were on any calcium channel blockers or beta-blockers during crizotinib treatment. The first patient was started on oral crizotinib 250 mg twice daily on November 17, 2008, and the last patients started crizotinib on June 6, 2011. The mean follow-up for all patients was 48 weeks (median, 42 weeks, 95% confidence interval, 4-116 weeks). Twenty-three patients still were receiving crizotinib treatment at the time of data cutoff. All patients had adenocarcinoma histology. None of the patients had dose reduction during crizotinib treatment. The clinicopathologic variables of all the patients are listed in Table 1. There was a significant difference in the mean age at diagnosis between patients who did and did not experience SB (P = .0366) (Table 1). There was a statistically significant difference in performance status between patients who did and did not experience SB (P = .0337), and 82.8% of patients who experienced SB, compared with only 46.2% of patients who did not experience SB, had a pretreatment performance status of 0 (Table 1). There was no difference between patients who did and did not experience SB according to sex (P = .7725), race (P = .8856), smoking status (P = .6678), or the number of prior regimens (P = .3298) (Table 1).
Table 1. Clinicopathologic Variables and Characteristics of Heart Rate Changes in 42 Patients Who Had ALK-Rearranged or MET-Amplified, Crizotinib-Treated Nonsmall Cell Lung Cancer Stratified by Those With and Without Sinus Bradycardia
|Patient characteristic|| || || || |
|Age, y|| || || || |
|Median [range]||51.0 [32-80]||52.0 [32-80]||48.0 [39-57]||.0336|
|Sex|| || || || |
|Men||24 (57.1)||17 (58.6)||7 (53.9)|| |
|Women||18 (42.9)||12 (41.4)||6 (46.1)||.7725|
|Ethnicity|| || || || |
|Caucasian||27 (64.3)||19 (65.5)||8 (61.5)|| |
|Asian||8 (19)||5 (17.2)||3 (23.1)|| |
|Hispanic||6 (14.3)||4 (13.8)||2 (15.4)|| |
|African American||1 (2.4)||1 (3.4)||1 (7.7)||.8856|
|Performance status|| || || || |
|0||30 (71.4)||24 (82.8)||6 (46.2)|| |
|1||11 (26.2)||5 (17.2)||6 (46.2)|| |
|2||1 (2.4)||0 (0)||1 (7.7)||.0337|
|Smoking status|| || || || |
|Never-smoker||38 (90.4)||26 (89.7)||12 (92.3)|| |
|Former/current smoker||4 (9.6)||3 (10.3)||1 (7.7)||.6678|
|Molecular alterations|| || || || |
|ALK-rearranged||41 (97.6)||28 (96.6)||13 (100)|| |
|MET-amplified||1 (2.4)||1 (3.4)||0 (0)||.9917|
|No. of previous regimens|| || || || |
|0-1||26 (61.9)||20 (69)||6 (46.2)||.3289|
|≥2||16 (38.1)||9 (31)||7 (53.8)|| |
|HR changes|| || || || |
|Mean pretreatment HR, bpm||84.9||77.9||100.6|| |
|Median [95% CI]||83 [60-115]||74 [61-103]||100 [80-137]||.0002|
|Mean lowest HR achieved, bpm||58.8||49.9||78.5|| |
|Median [95% CI]||54 [40-95]||52 [40-59]||75 [62-137]||< .0001|
|Mean maximum HR decrease, bpm||26.1||27.9||22.1|| |
|Median [95% CI]||26) [4-52]||27 [11-56]||24 [0-38]||.3685|
|Mean time to maximum HR decrease, wk||18.6||24.1||6.5|| |
|Median [95% CI]||9.5 [5-72]||14 [3-80]||6 [0-17]||.0089|
|Mean duration of treatment, wks||44.1||52.9||24.6|| |
|Median [95% CI]||28 [5-100]||44 [9-110]||19 [1-92]||.0050|
|Mean duration of response, wks||36.1||36.6||34.9|| |
|Median [95% CI]||26 [7-88]||27 [7-85]||21 [5-99]||.5222|
Heart Rate Changes Among All Patients and According to Sinus Bradycardia or No Sinus Bradycardia Experienced
Thirty-eight patients (90.4%) experienced at least 1 episode of an absolute decrease in HR of >10 bpm from pretreatment baseline. Twenty-nine patients (69) experienced at least 1 episode of SB (HR, <60). The mean maximum decrease in HR for all patients was 26.1 bpm (Table 1). There was a significant difference in the pretreatment HR (P = .0002), the mean lowest HR achieved (P < .0001), and the mean time to lowest HR recorded (P = .0089) between patients who did and did not experience SB (Table 1). It is noteworthy that there was no difference between the mean decrease in HR between patients who did and did not experience SB (P = .3685) (Table 1). There was a significant difference in the time to lowest HR recorded between patients who did and did not experience SB (P = .0089) (Table 1).
Profound Sinus Bradycardia
Among the 29 patients who experienced SB, 13 patients (44.8%) experienced at least 1 episode of profound SB (HR, <50 bpm). There was no difference in sex (P = .9476), race (P = .7791), performance status (P = .1411), or the number of prior regimens (P = .0902) among patients who experienced profound SB, SB, or no SB. Although there also was no difference in the age of diagnosis among the 3 groups, the patients who experienced profound SB had the oldest median age of diagnosis at 58.7 years followed by age 53.1 years for patients who experienced SB only and age 47.8 years for patients who experienced no SB (P = .0671).
There was a significant difference in the pretreatment HR (P = .0003), the mean lowest HR recorded (P < .0001), or mean time to maximum HR decrease (P = .0261) among patients who experienced profound SB, SB, and no SB (Table 2). It is worth noting that, again, there was no statistically significance difference in the average maximum HR decrease among patients who experienced profound SB, SB, and no SB (P = .5695) (Table 2). None of the patients who experienced SB or profound SB were symptomatic or had ECG changes, such as PR or QTc prolongation.
Table 2. Heart Rate Variables in Patients Without Sinus Bradycardia, With Sinus Bradycardia, and With Profound Sinus Bradycardia
|Mean pretreatment HR, bpm||100.6||81.5||74.0|| |
|Median (95% CI)||100.0 (80-137)||82.0 (60-115)||70.5 (61-103)||.0003|
|Mean lowest HR recorded, bpm||78.5||55.0||44.5|| |
|Median (95% CI)||75.0 (62-137)||55.0 (52-59)||44.5 (39-51)||< .0001|
|Mean time to maximum HR decrease, wk||6.5||18.7||29.9|| |
|Median (95% CI)||6.0 (0-17)||14.0 (3-80)||14.5 (2-120)||.0261|
|Mean maximum HR decrease, bpm||22.1||26.5||29.5|| |
|Median (95% CI)||24.0 (0-38)||30.0 (4-56)||26.0 (11-63)||.5695|
|Mean treatment duration, wk||24.6||47.7||58.4|| |
|Median (95% CI)||19.0 (1-92)||44.0 (8-116)||59.5 (9-110)||.0165|
|Mean response duration, wk||34.9||34.9||38.4|| |
|Median (95% CI)||21.0 (5-99)||27.0 (5-88)||30.5 (7-85)||.7292|
Clinical Response and Tumor Shrinkage
The overall response rate was significantly higher among patients who experienced SB than patients who experienced no SB (P = .0195) (Table 3). The mean maximum decrease in cumulative tumor measurement from baseline also was significantly higher among patients who experienced SB than those who did not (P = .0205) (Table 3). When only considering patients who had achieved clinical benefit (complete responses + partial responses + stable disease), the maximum tumor shrinkage also was numerically higher among patients who experienced SB than among those who did not, although the difference was not statistically significant (P = .0584) (Table 3).
Table 3. Clinical Response of the 42 Patients With ALK-Rearranged or MET-Amplified, Crizotinib-Treated Nonsmall Cell Lung Cancer Stratified According to Those With and Without Sinus Bradycardia
|All patients|| || || || |
|Best response to crizotinib: No. (%)|| || || || |
|CR||3 (7.1)||3 (10.3)||0 (0)|| |
|PR||18 (42.9)||15 (51.7)||3 (23.1)|| |
|SD||18 (42.9)||10 (34.5)||8 (61.5)|| |
|PD||1 (2.4)||1 (3.4)||0 (0)|| |
|Not evaluable||2 (4.8)||0 (0)||2 (15.4)||.0496|
|ORR: CR + PR, %||50||62.1||23.1||.0195|
|DCR: CR + PR + SD, %||92.9||96.6||84.6||.1650|
|Mean maximum decrease in cumulative tumor measurement from baseline for all patients, %||38.4||45.4||22.8|| |
|Median [95% CI]||35.5 [0-80]||53 [0-80]||21 [0-65]||.0205|
|Patients who achieved clinical benefit: CR + PR + SD|| || || || |
|No. of patients||39||28||11|| |
|Mean maximum decrease in cumulative tumor measurement from baseline for all patients, %||41.4||47||26.9|| |
|Median [95% CI]||39 [0-78]||53.5 [1-80]||29 [0-65]||.0584|
Similarly, the overall response rate was highest among patients who experienced profound SB, followed by patients who experienced SB, and then patients who experienced no SB (P = .0406) (Table 4). This also was similar in patients who experienced profound SB, who had the greatest reduction in cumulative tumor measurement from baseline, followed by patients who experienced SB, and then by patients who experienced no SB (P = .0447) (Table 4). Again, when only the patients who achieved clinical benefit from crizotinib were included, the cumulative tumor reduction was numerically highest among patients who experienced profound SB, although the difference was no longer significant (P = .1171) (Table 4).
Table 4. Clinical Response of Patients Who Had Crizotinib-Treated Nonsmall Cell Lung Cancer Without Sinus Bradycardia, With Sinus Bradycardia, and With Profound Sinus Bradycardia
|All patients|| || || || |
|Best response to crizotinib: No. (%)a|| || || || |
|CR||0 (0)||1 (6.7)||2 (14.3)|| |
|PR||3 (23.1)||7 (46.7)||8 (57.1)|| |
|SD||8 (61.5)||6 (40)||4 (28.6)|| |
|PD||0 (0)||1 (6.7)||0 (0)|| |
|Not evaluable||2 (15.4)||0 (0)||0 (0)||.1579|
|ORR: CR + PR, %||23.1||53.3||71.4||.0406|
|DCR: CR + PR + SD, %||84.6||93.3||100||.2992|
|Mean maximum decrease in tumor measurement, %||22.8||40.5||50.6|| |
|Median [95% CI]||21 [0-65]||39 [0-80]||56 [0-100]||.0447|
|Patients who achieved clinical benefit: CR + PR + SD|| || || || |
|No. of patients||11||14||14|| |
|Mean maximum decrease in tumor measurement, %||26.9||43.4||50.6|| |
|Median [95% CI]||29 [0-65]||46 [1-80]||56 [0-100]||.1171|
Heart Rate Changes and Clinical Response by Sex
There was no difference in the median age of diagnosis between men and women (men vs women: 53.9 years vs 52.6 years, respectively; P = .3597). There also was no difference in the mean treatment duration between men and women (50.5 weeks vs 35.7 weeks, respectively; P = .2970) or in the mean response duration (33.5 weeks vs 39.4 weeks, respectively; P = .4841). There was no difference in the mean pretreatment HR between men and women (83.6 bpm vs 86.7 bpm, respectively; P = .8189), the mean maximum decrease in HR recorded (27.8 bpm vs 23.9 bpm, respectively; P = .1814), the mean lowest HR achieved (55.8 bpm vs 62.8, respectively; P = .5330), or the mean time to the lowest HR recorded (22.1 weeks vs 14.1 weeks, respectively; P = .2684).
In terms of clinical response to crizotinib, there was no difference in the overall response rate between men and women (45.8% vs 55.6%, respectively; P = .5329). Similarly, there was no difference in the average maximum tumor decrease between men and women (35.8% vs 41.9%, respectively; P = .4450).
The major observation from this single-institution retrospective analysis of HR changes during crizotinib treatment is that HR decrease is a common phenomenon during such treatment. HR decrease has been reported as a pharmacodynamic phenomenon of crizotinib with an average of 2.5 bpm decrease per 100 ng of crizotinib. The average pharmacokinetic level of crizotinib achieved is approximately 280 ng/mL; thus, the average HR decrease should be approximately 8 bpm. We observed that 90% of patients who were receiving crizotinib had a 1-time HR decrease >10 bpm. The average HR decrease for all patients was approximately 26.1 bpm but without any symptoms or ECG changes (such as PR or QTc prolongation). The higher than expected HR decrease observed in this report may reflect individual pharmacokinetic variation or autoinhibition of crizotinib, because crizotinib is a strong cytochrome P450, family 3, subfamily A (CYP3A) inhibitor and also is itself metabolized by CYP3A.
Our exploratory analysis also revealed that patients who experienced SB had a significantly higher response rate and greater cumulative tumor shrinkage during crizotinib therapy, and patients who experience profound SB had the highest response rate and greatest cumulative tumor shrinkage. Patients who experienced profound SB also were on crizotinib treatment the longest. Thus, although this observation raises the possibility of a positive correlation between the magnitude of HR decrease and the extent of clinical benefit from crizotinib, it also may only reflect that the magnitude of HR decrease is a time-dependent or cumulative dose-dependent phenomenon. Nevertheless, it points to the importance for oncologists to try and avoid common medications that can cause HR lowering, such as beta-blockers and calcium channel blockers, during the entire course of crizotinib treatment, because patients can continue to experience HR decrease well into the course of their crizotinib treatment. We want to emphasize that most of our patients were not in permanent SB. Many of these patients alternated between normal sinus rhythm and SB during some of the subsequent clinic visits. Only a few patients were in persistent SB during the majority of clinic visits.
The second observation is that SB from crizotinib is more a reflection of the pretreatment HR rather than a true adverse event per se. We observed that the HR before patients received crizotinib treatment was significantly higher among patients who did not experience SB than among those who experienced SB, whereas there was no statistically significant difference in the average maximum HR decrease between these patient groups. We also observed that older patients and patients with good performance status (conditions that are generally associated with slower HR) were significantly more likely to experience SB while on crizotinib.
Neither CTCAE version 3 nor the newer version 4.0 specifies the HR at which it is considered SB. It is generally defined in cardiology that bradycardia occurs at an HR <60 bpm. However, an oncologist following a patient with ALK-rearranged NSCLC who is deriving clinical benefit from crizotinib and is asymptomatic may not consider an HR ≤59 bpm significant. Because CTCAE grades the severity of SB by the presence or absence of symptoms rather than by the actual lowest recorded HR, we may not appreciate the extent of HR decrease while reading the oncology literature. For example, our 3 previously reported asymptomatic patients who had HRs <45 bpm while on crizotinib all were graded as having grade 1 SB. In the future, we recommend incorporating specific ranges of the HR below 60 bpm in addition to symptoms experienced into the CTCAE grading of SB to allow a better appreciation of the extent of the SB by the oncology community.
It has been demonstrated that crizotinib causes a rapid but reversible drop in testosterone in men, which potentially may result in an HR decrease. Our analysis did not reveal any significant difference in the HR parameters examined and the clinical response to crizotinib between men and women. Thus, decrease in testosterone level is unlikely to be a mechanism that accounts for the HR decrease associated with the receipt of crizotinib. Another potential cause of HR decrease may be hypothyroidism; however, we did not measure thyroid hormone levels in any of our patients, because none of our patients exhibited any signs of hypothyroidism.
The mechanism(s) that allow crizotinib to slow down HR remain unknown. Given our observation that crizotinib lowers the HR without any drop in blood pressure, its effect on the heart is more likely to be chronotropic (which affects the sinoatrial lymph node) or dromotropic (which affects the atrioventricular lymph node) rather than inotropic. Alternatively, the bradycardiac effect of crizotinib may be because of its anti-MET effects, because an analysis of tivantinib, another MET inhibitor, also revealed HR decreases (Lee Rosen, University of California Los Angeles, personal communications).
With all of the limitations of this report, including its retrospective nature, the limited number of patients analyzed, and the recording of the HR at irregularly spaced intervals, depending on scheduled protocol visits, which varied and became less frequent with prolonged crizotinib treatment, it still represents the largest study to date describing HR changes during crizotinib treatment and exploring the potential relation between HR changes and clinical response to crizotinib. Furthermore, the data from this study have undergone extensive queries and quality-control by the clinical trial sponsor (Pfizer Inc., Groton, Conn), because they are part of the data submission to the US Food and Drug Administration in support the orphan drug application and eventual approval of crizotinib in the United States. Our exploratory analysis of a positive correlation of the magnitude of HR decrease with clinical response and tumor shrinkage should be considered only as hypothesis-generating. The PROFILE 1005 study has now enrolled more than 1000 patients with ALK-rearranged NSCLC and may provide a much more robust analysis of the frequency of, timing, patient characteristics associated with, and potential correlation to clinical response with HR changes during crizotinib treatment.
The PROFILE 1001 and 1005 trials were both funded by Pfizer.
CONFLICT OF INTEREST DISCLOSURES
Sai-Hong I. Ou has received honorarium from Pfizer.