Drug-induced reduction in estimated glomerular filtration rate in patients with ALK-positive non-small cell lung cancer treated with the ALK inhibitor crizotinib

Authors

  • Evelyn M. Brosnan MD, BS, MBA,

    1. Division of Medical Oncology, University of Colorado Anschutz Medical Campus, Aurora, Colorado
    Search for more papers by this author
  • Andrew J. Weickhardt MBBS, DMedSc,

    1. Division of Medical Oncology, University of Colorado Anschutz Medical Campus, Aurora, Colorado
    Search for more papers by this author
  • Xian Lu MSc,

    1. Department of Biostatistics and Informatics, Colorado School of Public Health, University of Colorado Anschutz Medical Campus, Aurora, Colorado
    Search for more papers by this author
  • Delee A. Maxon BS,

    1. Division of Medical Oncology, University of Colorado Anschutz Medical Campus, Aurora, Colorado
    Search for more papers by this author
  • Anna E. Barón PhD,

    1. Department of Biostatistics and Informatics, Colorado School of Public Health, University of Colorado Anschutz Medical Campus, Aurora, Colorado
    Search for more papers by this author
  • Michel Chonchol MD,

    1. Division of Renal Diseases and Hypertension, University of Colorado Anschutz Medical Campus, Aurora, Colorado
    Search for more papers by this author
  • D. Ross Camidge MD, PhD

    Corresponding author
    1. Division of Medical Oncology, University of Colorado Anschutz Medical Campus, Aurora, Colorado
    • Corresponding author: D. Ross Camidge, MD, PhD, Division of Medical Oncology, University of Colorado Cancer Center, University of Colorado Anschutz Medical Campus, 1665 North Aurora Ct, Room ACP 5327, Aurora, CO 80045; Fax: (720) 848-0536; ross.camidge@ucdenver.edu

    Search for more papers by this author

  • This study was successfully submitted as part of Dr. Brosnan's Capstone-mentored scholarly activity research project during her medical school education at the University of Colorado in Denver. Dr. Brosnan gratefully acknowledges the additional support and contributions of Dr. Jean Abbott, Lynne Fox, Dr. Rita Lee, and Dr. Stuart Linas to this project.

Abstract

BACKGROUND

To the best of the authors' knowledge, the renal side effects of crizotinib have not been investigated previously.

METHODS

The estimated glomerular filtration rate (eGFR) was calculated using the Chronic Kidney Disease Epidemiology Collaboration creatinine-based prediction equation during the first 12 weeks of crizotinib therapy and after crizotinib but before the introduction of any further systemic therapy.

RESULTS

A total of 38 patients with stage IV anaplastic lymphoma kinase (ALK)-positive non-small cell lung cancer who were treated with crizotinib were identified. The mean eGFR decreased by 23.9% compared with baseline (P < .0001; 95% confidence interval, 21.3%-26.6%), with the majority of the decrease occurring within the first 2 weeks of therapy. Clinical history and blood urea nitrogen/creatinine ratios did not suggest prerenal causes. The objective response rate among evaluable patients (n = 27) was 41%. Tumor shrinkage was not correlated with changes in eGFR (correlation coefficient, −0.052; P = .798). Among the 16 patients for whom data after treatment with crizotinib were available, recovery to within 84% of the baseline eGFR occurred in all patients. After adjusting for the number of scans with intravenous contrast and the use of known nephrotoxic drugs, the issue of whether a patient was on or off crizotinib treatment was found to be significantly associated with changes in eGFR (P < .0001).

CONCLUSIONS

As assessed by the Chronic Kidney Disease Epidemiology Collaboration prediction equation, eGFR is reduced by treatment with crizotinib, but the majority of patients will recover their eGFR after the cessation of therapy. The early onset, size of the change, minimal cumulative effect, and rapid reversibility raise the possibility that this may be a pharmacological and/or tubular creatinine secretion effect rather than a direct nephrotoxic effect. Increased vigilance with regard to the concomitant use of renally cleared medications or nephrotoxic agents should be considered for patients receiving crizotinib and, when eGFR is reduced, additional renal investigations should be undertaken. Cancer 2014;120:664–674. © 2013 American Cancer Society.

INTRODUCTION

Non-small cell lung cancer (NSCLC) comprises 84% of all lung cancers and carries a 5-year survival rate of 18% (for all disease stages).[1] Between 2% and 7% of NSCLC is driven by an anaplastic lymphoma kinase (ALK) fusion oncogene (ALK-positive NSCLC), which is detectable through a variety of assays.[2] In 2011, the US Food and Drug Administration granted accelerated drug approval to crizotinib (Pfizer Inc, New York, NY), a small-molecule inhibitor of ALK, based on the drug's ability to produce prolonged progression-free survival (PFS) and high objective response rates (ORRs) in patients with ALK-positive NSCLC.[2] In a follow-up phase 3 trial in patients with ALK-positive NSCLC, crizotinib was found to more than double PFS compared with standard chemotherapy and the drug was well tolerated, a finding that is consistent with previous studies.[3] Common adverse side effects reported in crizotinib trials include gastrointestinal disturbances (nausea, diarrhea, vomiting, and constipation), visual disturbances, and fatigue. The most common grade 3 to 4 side effects were neutropenia, transaminitis, hypophosphatemia, and lymphopenia.2-4 Since the approval of crizotinib, hypogonadism in males, asymptomatic bradycardia, and acute interstitial lung disease have also been noted.[6-8] Herein we describe what to our knowledge are the previously unreported side effects of a decreased estimated glomerular filtration rate (eGFR) associated with crizotinib use and explore its etiology in detail.[3, 4, 9]

MATERIALS AND METHODS

A retrospective review was conducted on a cohort of patients with stage IV ALK-positive NSCLC (AJCC 7th Edition) who were treated with crizotinib between 2009 and 2012. The data were retrieved from patient electronic medical records in accordance with a University of Colorado Institutional Review Board-approved protocol 09-018 and data analysis was performed up to December 31, 2012.

The eGFR was determined using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) creatinine-based equation, as shown in Table 1.

Table 1. eGFR as Determined Using the CKD-EPI Creatinine-Based Equation
SexSerum Creatinine, mg/dLCKD-EPI Equation
  1. Abbreviations: CKD-EPI, Chronic Kidney Disease Epidemiology Collaboration; eGFR, estimated glomerular filtration rate; SCr, serum creatinine.

Female≤0.7144 × (SCr/0.7)−0.329 × (0.993)age × [1.159 if African American]
Female>0.7144 × (SCr/0.7)−1.209 × (0.993)age × [1.159 if African American]
Male≤0.9144 x (SCr/0.7)−0.411 × (0.993)age × [1.159 if African American]
Male>0.9144 × (SCr/0.7)−1.209 × (0.993)age × [1.159 if African American]

This method was selected over the Modification of Diet in Renal Disease approach because CKD-EPI was developed from a study population that included both individuals without CKD as well as those with CKD. In studies, the CKD-EPI prediction equation had greater accuracy and less bias compared with Modification of Diet in Renal Disease for patients with a GFR > 60 mL/minute/1.73 m2, the measurement at which we expected our study population to be at baseline.[10, 11]

The median PFS with crizotinib therapy ranges from 7.7 months to 9.7 months and the majority of objective responses occur at the first assessment scan.[3, 4] To ensure adequate time for eGFR effects to manifest but maximize the size of the data set for analysis, including for correlations with tumor response, we focused on the first 12 weeks of crizotinib use. The percentage change in eGFR during the first 12 weeks of therapy with crizotinib was determined relative to baseline eGFR on the day of the initiation of crizotinib treatment, and the eGFR after crizotinib was determined compared with the eGFR on the last day of crizotinib treatment.

Each record was examined for the use of drugs or other agents that could alter renal function, including nonsteroidal antiinflammatory drugs (NSAIDS), angiotensin-converting enzyme inhibitors, angiotensin receptor blockers, diuretics, metformin, prior cisplatin/carboplatin, and the use of intravenous (iv) contrast (excluding fluorodeoxyglucose only) for imaging. Preexisting diabetes, hypertension, and hyperlipidemia were noted. Proteinuria was assessed using available urine dipstick results. Systolic blood pressure data and any documented emesis or diarrhea as graded according to version 3.0 of the National Cancer Institute Common Terminology Criteria for Adverse Events were extracted from treatment visit notes. The ratios of blood urea nitrogen to serum creatinine (BUN/Cr) were calculated for available data in the first 2 weeks of therapy.

Statistical Analysis

The association between tumor shrinkage (determined according to Response Evaluation Criteria In Solid Tumors [RECIST] version 1.1) at the first imaging scan and changes in the eGFR from baseline at the closest time point to that imaging scan was examined using a Spearman correlation.

All statistical analyses were performed using SAS/STAT statistical software (version 9.3; SAS Institute Inc, Cary, NC). Longitudinal analyses of eGFR values were performed using piecewise linear mixed models.[12] These models accounted for the correlation of repeated observations within a patient and allowed for a possible lack of balance in the number and timing of measurements over time. Each piece in the model was defined by a time point (a “knot”) on either side of it, with knots at 2, 4, 6, 8, and 12 weeks for which creatinine data were available and a normalized eGFR could be calculated. Between each set of 2 consecutive knots, a regression line was fitted (a “spline”) to determine the slope and thus the direction of the change in eGFR. The mean difference in eGFR between the spline's beginning and ending knots was determined using linear contrast (ie, week 2 vs baseline, week 4 vs week 2, etc). Finally, P values were calculated for each eGFR knot relative to baseline and relative to the previous time piece to assess whether a difference in the eGFR was statistically significant.

A subgroup of patients in the cohort was also analyzed for whom data regarding time spent “on/off” crizotinib were available. In this cohort, ongoing clinical and laboratory data were available after discontinuation of crizotinib dosing due to either toxicity or disease progression but before the initiation of any additional systemic anticancer therapy. In this subgroup, the percentage change in the eGFR relative to the last day of crizotinib treatment at the maximum time after crizotinib free of any subsequent systemic therapy was compared with the percentage change in the eGFR compared with the baseline value before crizotinib at the closest time-matched period on crizotinib. For the subset of patients who went off therapy with crizotinib, a separate statistical model was fitted and corresponding estimates of changes in the eGFR were obtained. To test the potential confounding effects of iv contrast agents and other medications, matched outcome data for the eGFR while on and off crizotinib therapy were compared using a linear mixed-effects model with a random patient effect and a fixed period effect while controlling for the number of scans and other medications.

RESULTS

A total of 38 patients with stage IV ALK-positive NSCLC were identified. As outlined in Figure 1, 27 of these patients had measurable disease while receiving therapy and 16 had “on/off” crizotinib data available. The demographics of these 2 cohorts are listed in Tables 2 and 3. All patients commenced crizotinib treatment orally at the standard dose of 250 mg twice daily. The median duration of crizotinib treatment was 16.3 months (range, 1 month-39 months). One patient with baseline severe renal failure (eGFR of 35.0 mL/minute/1.73 m2) secondary to prior cisplatin therapy experienced several treatment interruptions and dose reductions due to worsening kidney function before eventually discontinuing the drug in the absence of disease progression (Fig. 2). As data on the detailed effects of crizotinib on the eGFR emerged, this patient was rechallenged with additional measurements of renal function conducted prospectively before and after recommencement of crizotinib (Fig. 2). Two other patients required dose reductions in crizotinib due to neutropenia (down to 200 mg twice daily and 150 mg twice daily, respectively), whereas a third patient required a dose reduction down to 200 mg twice daily due to transaminitis. In each case, the reduced dose was maintained for the remainder of their treatment. All these patients had received their first tumor assessment scan before the dose reductions, except for the patient who experienced transaminitis. For this patient, the dose was reduced after the first cycle and occurred before the first assessment scan.

Figure 1.

Breakdown of the cohort of 38 patients is shown. CZ indicates crizotinib; CT, computed tomography; RECIST, Response Evaluation Criteria In Solid Tumors.

Table 2. Baseline Patient Demographics (n=38){TC}
CharacteristicNo.
  1. Abbreviations: ACEI, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; DM, diabetes mellitus; eGFR, estimated glomerular filtration rate; HLD, hyperlipidemia; HTN, hypertension; NSAIDs, nonsteroidal antiinflammatory drugs; UA, xxx.

Total38
No. of women18
Median age (range), y54.5 (28.8-71.2)
No. of men20
Median age (range), y54.9 (22.9-79.8)
Race 
White33
Hispanic3
African American1
Asian1
Median duration of crizotinib (range), mo16.3 (1-39)
No. with dose interruption5
No. with dose reduction4
Baseline eGFR, mL/minute/1.73 m2 
Mean82.6
Median78.4
Range35.0-127.3
No. with baseline UA with proteinuria (amount)6 (trace-30 mg/dL)
No. with preexisting conditions 
DM2
HTN5
HLD4
Prior use of drugs 
NSAIDS (occasional use)11
ACEIs1
ARBs0
Diuretics0
Metformin0
Cisplatin (range in mo prior to crizotinib)14 (4-60)
Carboplatin (range in mo prior to crizotinib)28 (4-48)
Table 3. Patient Demographics for the On/Off Crizotinib Subgroup (n=16){TC}
CharacteristicNo.
  1. Abbreviations: ACEI, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; DM, diabetes mellitus; eGFR, estimated glomerular filtration rate; HLD, hyperlipidemia; HTN, hypertension; NSAIDs, nonsteroidal antiinflammatory drugs; UA, xxx.

Total no.16
No. of women9
Median age (range), y60.0 (22.8-71.2)
No. of men7
Median age (range), y48.8 (34.5-56.5)
Race 
White15
Hispanic1
Median duration of crizotinib (range), mo17.8 (4.8-34.4)
No. with dose interruption3
No. with dose reduction2
Baseline eGFR, mL/minute/1.73 m2 
Mean84.4
Median85.3
Range35.0-127.3
No. with baseline UA with proteinuria (amount)6 (trace-30 mg/dL)
No. with preexisting conditions 
DM0
HTN2
HLD0
Prior use of drugs 
NSAIDS (occasional use)6
ACEIs0
ARBs0
Diuretic (furosemide)0
Cisplatin (range in mo prior to crizotinib)5 (10-60)
Carboplatin (range in mo prior to crizotinib)13 (4-56)
Table 4. Time-Matched Comparison On/Off Crizotinib (n=16)
Patient No.Time Off CZ and All Systemic Therapy, WeeksNo. of Scans With Contrast in the Time Period Off TreatmentMaximum % Change in eGFR During Off PeriodaDrugs Received During Off PeriodMatched Time Period From Start On CZ, WeeksNo. of Scans With Contrast in Matched Time Period On TreatmentMaximum % Change in eGFR During Time-Matched Period On TreatmentDrugs Received During Matched Period On Treatment
  1. Abbreviations: CZ, crizotinib; eGFR, estimated glomerular filtration rate; NSAIDS, nonsteroidal antiinflammatory drugs; prn, as needed.

  2. a

    GFR normalized to that when therapy with CZ was stopped.

1122113.6NSAIDS prn123−32.07NSAIDS prn
28258.61None82−15.57None
34313.16NSAIDS prn40−20.94NSAIDS prn
42229.61NSAIDS prn20−28.21NSAIDS prn
53033.32Furosemide40−22.47Furosemide
61018.01NSAIDS prn203.93Furosemide
74425.91None40−15.37None
10103.67None20−10.1None
152044.73NSAIDS prn20−16.45NSAIDS prn
2210−1.32None20−22.4None
2412645.34NSAIDS prn122−42.2NSAIDS prn
265236.5Furosemide40−9.12None
282025.12None20−19.6None
292211.24NSAIDS prn20−7.14None
3010−5.9None102.63None
38928.93None92−34.28None
Figure 2.

(A) Serum creatinine readings (in mg/dL) are shown in a single patient with preexisting renal impairment secondary to cisplatin exposure demonstrating initial elevations during crizotinib exposure and reelevation on reexposure 17 months later. The patient's tumor responded to the crizotinib each time. (B) Expansion of the area from panel A is shown, demonstrating details of dose modifications tried during the initial crizotinib exposure. Each cycle was 21 days in length. The blue and green lines represent the upper and lower limits of the normal range, respectively. Rx indicates treatment; C1 Day 15, Cycle/Day 15 (28 day cycle); BID, twice daily; QD, every day; Cr, creatinine.

Consistent with our selection of the CKD-EPI creatinine-based prediction equation for the eGFR, the mean baseline eGFR for the cohort of 38 patients was 82.6 mL/minute/1.73 m2 and the median eGFR was 78.4 mL/minute/1.73 m,[2] with a range of 35.0 mL/minute/1.73 m2 to 127.3 mL/minute/1.73 m2. Four patients had an eGFR ≤ 60 mL/minute/1.73 m2 (35.0 mL/minute/1.73 m2, 45.9 mL/minute/1.73 m2, 47.0 mL/minute/1.73 m2, and 60.0 mL/minute/1.73 m2, respectively). In the subgroup of 16 patients in whom on/off crizotinib time periods were compared, the mean baseline eGFR was 84.4 mL/minute/1.73 m2 and the median was 85.3 mL/minute/1.73 m2, with a range of 35.0 mL/minute/1.73 m2 to 127.3 mL/minute/1.73 m2. Two of these patients had an eGFR ≤ 60 mL/minute/1.73 m2 (35.0 mL/minute/1.73 m2and 47.0 mL/minute/1.73 m2, respectively).

Creatinine-Derived eGFR Decreases With Crizotinib Therapy

For all 38 patients, there was a mean 23.9% decrease noted in the CKD-EPI eGFR over the first 12 weeks of treatment with crizotinib (P < .0001; 95% confidence interval [95% CI], 21.3%-26.6%). The reduction in eGFR was steepest in the first 2 weeks, with a mean decrease of 19.9% (P < .0001; 95% CI, 14.4%-25.4%) as shown in Figure 3.

Figure 3.

Percentage change in the creatinine-derived estimated glomerular filtration rate (eGFR) from baseline during first 12 weeks of crizotinib therapy is shown (individual patient data; n = 38). Pt indicates patient.

Dehydration or Additional Drug Use Does Not Explain the Steep Drop in the eGFR Noted Within the First 2 Weeks

Patients reported no significant gastrointestinal issues during the first 2 weeks that would lead to volume depletion causing an acute kidney injury. None of the 38 patients experienced emesis during this time and 4 patients reported only grade 1 intermittent diarrhea that was controlled with loperamide hydrochloride. Only 1 patient in the cohort used diuretics during this period (furosemide at a dose of 40 mg daily). At the 2-week clinic visit, none of the patients had hypotensive blood pressure measurements and none had BUN/Cr ratios > 20:1 to suggest prerenal causes for the impaired eGFR (mean BUN/Cr of 12.0, median of 11.6 [range, 6.6-18.9])

Tumor Lysis Does Not Account for the Change in the eGFR Noted During Crizotinib Therapy

Consistent with prior reports of crizotinib's activity, there was a 41% ORR in the 27 patients in the cohort with evaluable disease at the time of the first assessment scan (mean, 7.4 weeks; median, 7.3 weeks [range, 4 weeks-15 weeks] after the initiation of crizotinib).[2-4] No correlation was found between percentage tumor shrinkage at the first assessment scan and the decrease in the eGFR at the closest time point to the radiographic assessment in these 27 patients (correlation efficient [r], −0.052; P = .798), as illustrated in Figure 4.

Figure 4.

Correlation between percentage change in tumor size at the time of first radiographic assessment and change in the estimated glomerular filtration rate (eGFR) at an equivalent time point (n = 27) is shown. Recist indicates Response Evaluation Criteria In Solid Tumors.

Crizotinib Exposure Appears to Be Directly Associated With the Change in the eGFR

In the on/off cohort (Table 4), a mean 15.9% drop in the eGFR as determined by the CKD-EPI equation occurred during the first 12 weeks of treatment with crizotinib (P < .0001; 95%CI, 13.4%-18.4%). Similar to the overall cohort, the greatest decrease occurred during the first 2 weeks of therapy (mean, 15.8%; P = .0001 [95% CI, 9.3%-22.3%]).

To explore possible confounders, this cohort was further analyzed by comparing the eGFR at the maximum time period that the patient was off crizotinib, and any other chemotherapeutic drugs, with the equivalent time period after starting the drug initially (the “time-matched periods”). The median time off crizotinib in these patients was 2.5 weeks (range, 1 week-12 weeks). There were no significant differences in the exposure to iv contrast, NSAIDs, or diuretics noted in the patients between the on/off assessment periods (P = .325, P = .371, and P = .587, respectively). No angiotensin-converting enzyme inhibitors were used in these patients during either the on or off crizotinib analysis periods.

At the time of ceasing treatment with crizotinib, the eGFRs of the patients demonstrated an average 35.3% recovery (95% CI, 15.2%-55.4%) from the last treatment day at the maximum time assessed off crizotinib (P = .0018), as shown in Figure 5. At the equivalent time-matched period on crizotinib, the mean decrease in the eGFR compared with baseline was 15.2% (95% CI, 6.9%-23.5%). In this cohort, the eGFR was on average 50.5% lower during the on-drug period compared with the off-drug period (P < 0.0001). After adjusting for the number of scans with iv contrast and the use of drugs that could alter renal function, whether a patient was on or off crizotinib remained highly statistically significantly associated with changes in the eGFR (P < .0001).

Figure 5.

Recovery of the estimated glomerular filtration rate (eGFR) at the maximum time after crizotinib treatment (compared with the last day of treatment) and change in the eGFR (compared with baseline) at an equivalent time-matched point during treatment with crizotinib are shown (individual patient data; n = 16).

Recovery of the eGFR Can Be Complete

When the eGFR at the maximum time off crizotinib was normalized to the original baseline value (as opposed to that on the last day of treatment), the mean eGFR increased 18.9% over baseline (P = .047; 95% CI, 0.3%-37.4%), ranging from −16.0% to 84.2%. Recovery of the eGFR to baseline levels or above occurred in 56.3% of patients and the remaining 43.8% of patients recovered to within 84% to 97% of their baseline eGFR (Fig. 6). Advanced patient age did not appear to be a factor in the recovery because the median age among those patients with complete recovery of their baseline eGFR was 60 years, whereas among those patients with only partial recovery the median age was 45.7 years. Baseline renal impairment also did not limit recovery. In fact, using a Spearman correlation test, patients' recovery (represented by the maximum percentage change in the eGFR after crizotinib compared with their baseline value) was inversely correlated with their baseline eGFR measurements (r, −0.68; P = .0036).

Figure 6.

Recovery of the estimated glomerular filtration rate (eGFR) compared with baseline after treatment with crizotinib was withdrawn is shown (individual patient data; n = 16). Pt indicates patient.

Prospective Renal Measurements Before and During Crizotinib Therapy

A white man aged 58 years with stage IV ALK-positive NSCLC and preexisting renal impairment secondary to prior cisplatin exposure was initially deemed intolerant of crizotinib due to recurrent elevations in serum creatinine associated with its use (Fig. 2). Due to increased awareness of the potential reversibility and limited cumulative effect of crizotinib on renal function generated from the data set presented herein, approximately 17 months later it was considered safe to cautiously rechallenge this patient with crizotinib at a dose of 250 mg twice daily. On May 31, 2013, before crizotinib was initiated on June 3, 2013, his serum creatinine was 2.09 mg/dL, his eGFR as calculated using the CKD-EPI equation was 34 mL/minute/1.73 m2, and his creatinine clearance measured through a 24-hour urine collection was 47.38 mL/minute/1.73 m2 (1426 mg creatinine/day in urine). On June 17, 2013, after having received crizotinib for 15 days, his serum creatinine was 2.64 mg/dL (26% increase), his eGFR as calculated using the CKD-EPI equation was 26 mL/minute/1.73 m2 (24% decrease), and his creatinine clearance as measured through a 24-hour urine collection was 34.90 mL/minute/1.73 m2 (26% decrease) (1307 mg creatinine/day in urine, an 8% decrease). Urine microscopy was unremarkable and did not demonstrate any evidence of acute tubular necrosis (ie, granular casts or renal tubular epithelial cells). He achieved a complete metabolic response on his first positron emission tomography/computed tomography scan performed on July 23, 2013 and was continuing to receive crizotinib at the time of last follow-up. His last creatinine level, measured on July 24, 2013, was stably elevated at 2.67 mg/dL.

DISCUSSION

Crizotinib is a potent inhibitor of several different oncogenic kinases including ALK, ROS1, and MET.[4, 13, 14] Crizotinib is a small molecule and is predominantly eliminated in the feces (63%) and urine (22%).[9] Its use in patients with ALK-positive NSCLC is associated with ORRs of 57% to 65% and a median PFS of between 7.7 months and 9.7 months.[2-4] Because disease progression during crizotinib therapy can occur in either the central nervous system through inadequate drug exposures or in the rest of the body due to the selection of specific resistant clones, local treatment of isolated areas of disease progression and the continued use of crizotinib for many months after disease progression has also recently been described.[15, 16] Consequently, the median duration of crizotinib exposure in many patients with ALK-positive disease may be considerably longer than the median PFS described within the initial studies.

In the current study, we described a previously unreported side effect of crizotinib in relation to the drug decreasing the eGFR as assessed by the CKD-EPI creatinine-based prediction equation. The mean eGFR was reduced by 23.9% over the first 12 weeks of dosing with crizotinib (P < .0001; 95% CI, 21.3%-26.6%). Because of the routine monitoring of creatinine in these patients, we were able to track the timescale of this effect. The greatest reduction in the eGFR occurred during the first 2 weeks of therapy when the mean decrease in the eGFR was 19.9% compared with baseline readings (P < .0001; 95% CI, 14.4%-25.4%). Cumulative effects beyond the initial rapid reduction were rare (Fig. 3).

The current study was limited by its retrospective design and the small sample size available for analysis, and therefore several caveats need to be considered. Although the rapidity and consistency of the effect strongly implicated crizotinib as the cause, other confounders have to be ruled out. Patients with disease progression may lose weight, or patients who are responding or developing peripheral edema (a known side effect of crizotinib) may gain weight; however, weight is not involved in the CKD-EPI calculation of GFR and therefore any such changes should not have affected our eGFR readouts within the current study. Patients did not have significant episodes of hypotension, emesis, or diarrhea documented during the time periods analyzed to account for the observed renal impairment, and the BUN/Cr ratios did not suggest a prerenal cause. Crizotinib is an active drug, producing a 41% ORR within our response-evaluable cohort at the time of the first assessment scan. Although the majority of objective responses on crizotinib occur at the time of the first assessment, later responses also can occur.[2, 4] Allowing for this, the activity of crizotinib in the current study cohort appears comparable to that noted within the literature. Although tumor lysis in highly treatment-responsive malignancies can be associated with acute kidney injury, we found no correlation between the percentage of tumor shrinkage as per RECIST and the amount of change in the eGFR at the closest time point to the radiographic assessment (r, −0.052; P = .798) (Fig. 4).[17, 18] Within a subgroup of patients in whom we could capture eGFR changes both on crizotinib and after a period of time after therapy with crizotinib but before the introduction of any other systemic anticancer therapy, we were able to isolate the effect of crizotinib separate from exposure to other potential nephrotoxic agents including NSAIDs, diuretics, and iv contrast from imaging studies (Table 4). After treatment with crizotinib, the eGFR increased in 14 of 16 patients compared with the values on the last day of crizotinib treatment (Fig. 5). Using a linear mixed-effects model, the issue of whether a patient was on or off crizotinib therapy was found to be highly statistically significantly associated with the noted changes in the eGFR as determined by the CKD-EPI equation (P < .0001). The majority of patients start to experience improvements in their eGFR within weeks of stopping crizotinib and this recovery is complete in 56.3% of cases; for the remaining 43.8% of cases, the eGFR returns to within 84% to 97% of the starting eGFR (Fig. 6). Because the duration of time off crizotinib in this cohort was relatively short, with a median of 2.5 weeks (range, 1 week-12 weeks), and the patients were likely to have had progressive disease and may have been deteriorating physically as well, these data may, if anything, underestimate the extent of eGFR recovery that is possible once patients are taken off crizotinib therapy.

In the current study cohort, the median duration of therapy was 16.3 months (range, 1 month-39 months), but we focused only on the first 12 weeks of therapy. Therefore, the question of whether the eGFR changes during therapy would be more pronounced if we examined changes over a longer period of time has to be raised. Certainly, the maximum decrease in the eGFR at any time during therapy was higher than that during the first 12 weeks in some cases (data not shown). However, such data could be biased by, for example, the intermittent use of diuretics to address late-onset peripheral edema rather than a true cumulative effect.[4] In addition, because longitudinal analyses using piecewise linear mixed models demonstrated that the majority of the eGFR changes occur within the first 2 weeks of therapy and then plateau, most of the relevant clinical data are likely to be contained within our presented analyses (Fig. 3).

In terms of understanding the clinical significance of these findings, we need to consider the basis of the eGFR calculation within this study. Specifically, it is important to remember that, although creatinine is mainly cleared by glomerular filtration, 10% to 20% is also secreted by the proximal tubule via the organic cation transporter 2 in the proximal tubule of the kidney.[19-21] The mechanism for the decrease in the eGFR (as calculated by a rise in creatinine) while crizotinib is being administered is unclear from the current study. However, given that the mean decrease in the eGFR was 20% to 25% and that it occurred rapidly and then plateaued and was largely reversible on cessation of dosing, direct nephrotoxicity appears to be unlikely. Instead, it is tempting to speculate that the effect either reflects direct inhibition of an as-yet unknown kinase involved in renal function and/or that this effect is due to interference with the tubular secretion of creatinine. With regard to creatinine secretion, many drugs are also substrates for organic cation transporters and have been shown to act as competitive inhibitors of creatinine at organic cation transporters, causing elevated serum creatinine levels. For example, cimetidine has been reported to cause an increase in creatinine of 20% to 30%, without a change in the true GFR as measured by inulin clearance.[22, 23] Other studies have found similar effects with both trimethoprim and pyrimethamine.[24-27] Perhaps crizotinib might also act as a competitive inhibitor at the creatinine transporter. In the patient described herein, in whom we documented 24-hour urine collections before and after the administration of crizotinib, the decrease in urinary creatinine (8%) was below the mean reduction described in a previous detailed cimetidine study (14%).[28] However, it was still within the range of urinary creatinine reductions reported (range, 0.3%-35%).[28] Unfortunately, we do not have paired data from direct measurements of the GFR (eg, inulin studies or iothalamate nuclear medicine scans) either before and during or during and after treatment with crizotinib to compare with our creatinine-based analyses to definitively prove, or disprove, an effect of crizotinib on the tubular secretion of creatinine. Without this information, an effect on tubular creatinine secretion represents only one plausible hypothesis.

Patients may remain on crizotinib for months or even years, including ongoing use of the drug after the initial instance of disease progression.[2, 4, 16] Consequently, although its uncertain etiology and reversibility may justify cautiously continuing crizotinib in patients who are deriving benefit from the drug and in whom creatinine levels rise significantly outside of the normal range, increased vigilance with regard to the concomitant use of known nephrotoxic or renally cleared drugs in patients receiving crizotinib is currently warranted. Because the true mechanism underlying crizotinib's effects on the creatinine-based eGFR remains unknown, when the eGFR is reduced and clinical decisions are being made based on these results, additional renal investigations to further define the true GFR in patients should be considered.

FUNDING SUPPORT

Statistical support was provided by the biostatistics core of the University of Colorado Specialized Program of Research Excellence (SPORE) in lung cancer (grant P50CA058187). Molecular analyses were conducted in part with the assistance of the Molecular Pathology Shared Resource of the University of Colorado Comprehensive Cancer Center (grant P30CA046934).

CONFLICT OF INTEREST DISCLOSURES

Dr Weickhardt has received a previous honorarium from Pfizer. Dr. Camidge has acted as a paid member of the Advisory Board and/or as a consultant for Servier, Eli Lilly, Genentech/Roche, Astex Pharmaceuticals, ImmunoGen, Clarient, Exelixis, IndiPharm, Astellas, Chugai, Clovis, AstraZeneca, Aveo, and Novartis. He has also received fees as a member of the Advisory Board and/or as a consultant for, for seminars and talks to members of the industry from, and research funding from Ariad and has acted as a paid member of the Advisory Board and/or as a consultant for and received fees for seminars and talks to members of the industry from Boehringer Ingelheim, Array BioPharma, Synta Pharmaceuticals Corporation, and Pfizer.

Ancillary