Brief Report: Increased Hepatotoxicity Associated with Sequential Immune Checkpoint Inhibitor and Crizotinib Therapy in Patients with Non-Small-Cell Lung Cancer

Abstract

Introduction:

Immune checkpoint inhibitors (ICIs) are standard therapies in advanced non-small-cell lung cancer (NSCLC). While genotype-directed tyrosine kinase inhibitors (TKIs) represent the standard of care for subsets of oncogene-driven NSCLC, patients may receive ICIs during their disease course. The impact of sequential ICI/TKI therapy on the risk of hepatotoxicity has not been described.

Methods:

Patients with advanced ALK/ROS1/MET-driven NSCLC treated with crizotinib, with or without preceding ICI, were identified. The cumulative incidences of crizotinib-associated grade 3+ transaminase elevations (per the Common Terminology Criteria for Adverse Events version 4.0) were compared.

Results:

We identified 453 patients with NSCLC harboring an oncogenic alteration in ALK, ROS1, or MET who were treated with crizotinib (11 with, 442 without prior ICI). Among 11 patients treated with ICI followed by crizotinib, five (cumulative incidence, 45.5%; 95% confidence interval [CI], 14.9-72.2) developed grade 3 or 4 ALT elevation, and four (36.4%; 95% CI, 10.0-64.2) developed grade 3 or 4 AST elevation. In comparison, among 442 patients who received crizotinib only, grade 3 or 4 ALT and AST elevations occurred in 34 (8.1%; 95% CI, 5.7-11.0; P <0.0001) and 14 patients (3.4%; 95% CI, 1.9-5.5; P <0.0001), respectively. There were no grade 5 transaminitis events. All cases of hepatotoxicity following sequential ICI and crizotinib use were reversible and nonfatal, and no case met Hy’s law criteria.

Conclusions:

Sequential ICI/crizotinib treatment is associated with a significantly increased risk of hepatotoxicity. Careful consideration and monitoring for hepatotoxicity may be warranted in patients treated with crizotinib following prior ICI.

INTRODUCTION

Immune checkpoint inhibitors (ICIs) targeting programmed cell death 1 (PD-1) and its ligand PD-L1 have entered the clinic across multiple solid tumors, including non-small-cell lung cancer (NSCLC).¹ Based on KEYNOTE-189, pembrolizumab plus platinum-doublet chemotherapy is the standard first-line treatment for nonsquamous NSCLC lacking EGFR mutation or ALK rearrangement regardless of PD-L1 status,² and pembrolizumab alone represents an additional first-line option for NSCLC with high-level PD-L1.³ In clinic, PD-L1 results may return faster than other genetic testing (e.g., EGFR, ALK, ROS1, and BRAF alterations), or tissue may be insufficient for genotyping, prompting the initiation of an ICI before genotype-directed tyrosine kinase inhibitors (TKIs).

Thus far, insights into the potential toxicities of sequential ICI and TKI therapy are limited. In one observational study, the risk of EGFR TKI-associated pneumonitis was substantially higher with sequential or concurrent ICI/TKI, although other toxicities were not investigated.⁴ In a separate phase II study of pembrolizumab in TKI-naïve EGFR-mutant NSCLC, 2 of 7 patients subsequently treated with an EGFR TKI experienced grade 3 transaminitis and grade 5 pneumonitis, respectively, similarly raising concerns of increased risks with sequential ICI and TKI.⁵ Notably, a high incidence of hepatotoxicity has been reported with certain concurrent ICI/TKI strategies in NSCLC.^6-8 For example, a phase 1/2 study of nivolumab plus crizotinib in ALK-rearranged NSCLC was terminated early due to severe hepatotoxicity in 5 of 13 (38%) patients.⁶ Two patients died, highlighting the importance of diagnosing and intervening on potentially fatal hepatotoxicity. To our knowledge, no study has evaluated whether ICI followed by TKI use is associated with an elevated risk of hepatotoxicity. ICIs and TKIs individually can cause hepatotoxicity, and ICIs have long half-lives which could impact the development of subsequent TKI-associated hepatotoxicity.

Crizotinib is a multitargeted ALK/ROS1/MET inhibitor, which remains the only approved, standard firstline therapy for ROS1-rearranged NSCLC.⁹ It also represents a clinically meaningful treatment option for NSCLC harboring MET alterations and recently received FDA breakthrough therapy designation for NSCLC with MET exon 14 skipping.¹⁰ Here, we performed a retrospective study of 453 patients with ALK/ROS1/MET-altered advanced NSCLC who received crizotinib with or without prior ICI, in order to address whether sequential ICI/crizotinib use increases hepatotoxicity risk.

METHODS

We identified patients with advanced NSCLC harboring an ALK fusion, ROS1 fusion, MET exon 14 skipping or amplification, and treated at Massachusetts General Hospital between 1/2008 and 4/2018. Sequential treatment with an ICI followed by crizotinib was required. For the comparator, we identified patients treated with crizotinib only. Hepatotoxicity occurring during crizotinib treatment was annotated and graded per the Common Terminology Criteria for Adverse Events version 4.0. All data were updated as of April 1, 2018. Records were reviewed under an institutional review board-approved protocol.

Categorical and continuous variables between groups were compared using Fisher’s exact test and Wilcoxon rank-sum test, respectively. The cumulative incidence of hepatotoxicity was estimated from crizotinib start to the onset of grade 3 or 4 transaminitis, and compared between groups using Gray’s test. Among patients treated with crizotinib who did not experience hepatotoxicity, the date of crizotinib discontinuation was analyzed as a competing risk, or censored on the last follow-up date if continuing on crizotinib at the time of data analysis. Statistical analysis was performed using Stata version 14.2 and SAS v9.4, with p-values based on a two-sided hypothesis.

RESULTS

We identified 453 patients with ALK (n=345), ROS1 (n=83), or MET (n=25) driven NSCLC. Eleven received an ICI prior to crizotinib, and 442 received crizotinib only (Supplemental Table 1). Baseline patient characteristics between the sequential ICI/crizotinib and crizotinib-only cohorts were overall comparable; however, more patients in the sequential ICI/crizotinib cohort had MET-driven NSCLC and smoking history (Supplemental Table 1). There was no difference in the presence of liver conditions [0/11 (0%) versus 16/442 (4%), P >0.999] or hepatic metastases prior to crizotinib [2/11 (18%) versus 101/442 (23%), P >0.999] between the cohorts (Supplemental Table 1).

Five of 11 (cumulative incidence 45.5%; 95% CI, 14.9-72.2) patients receiving sequential ICI/crizotinib developed grade 3/4 ALT elevation; three (27.3%) were grade 4. In comparison, 34 of 442 (8.1%; 95% CI, 5.4-10.5; P <0.0001) patients treated with crizotinib only developed grade 3/4 ALT elevation, of which four (0.9%) were grade 4. Grade 3/4 AST elevation occurred in four of 11 (cumulative incidence 36.4%; 95% CI, 10.0-64.2) patients receiving sequential ICI/crizotinib, versus 14 of 442 (3.4%; 95% CI, 1.9-5.5; P <0.0001) receiving crizotinib only (Table 2). Grade 5 hepatotoxicity was not observed in either cohort. The median interval between crizotinib initiation and onset of grade 3+ transaminitis was 38 days (range, 13-49) versus 41 days (range, 5-1657) for sequential ICI/crizotinib and crizotinib-only patients, respectively. The median interval between the last ICI dose and crizotinib initiation was 30 days (range, 21-135) among the sequential ICI/crizotinib cohort patients who developed grade 3/4 transaminase elevation, versus 50 days (range, 10-121) among those who did not.

Table 2.

ALT/AST elevation with crizotinib following ICI versus crizotinib alone.

	Patients, No.		Cumulative incidence of liver toxicity
	Total	Liver toxicity	Point estimate (%)	(95% CI)	P value
Grade 3/4 ALT elevation					<0.0001
ICI → TKI	11	5	45.5	(14.9, 72.2)
TKI	442	34	8.1	(5.7, 11.0)
Grade 4 ALT elevation					<0.0001
ICI → TKI	11	3	27.3	(5.8, 55.4)
TKI	442	4	0.9	(0.3, 2.2)
Grade 3/4 AST elevation					<0.0001
ICI → TKI	11	4	36.4	(10.0, 64.2)
TKI	442	14	3.4	(1.9, 5.5)
Grade 4 AST elevation					<0.0001
ICI → TKI	11	3	27.3	(5.8, 55.4)
TKI	442	1	0.2	(0.02, 1.3)

Grading is per the Common Terminology Criteria for Adverse Events version 4.0.

Point estimate is reported at the time of last observed event.

Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase; TKI, tyrosine kinase inhibitor; ICI, immune checkpoint inhibitor; CI, confidence interval.

Of note, among five patients who received sequential ICI/crizotinib and developed grade 3+ transaminitis, none had pre-existing autoimmune or liver conditions. One (patient 3 in Table 1) had hepatic metastases with baseline grade 1 transaminitis; all others had normal baseline transaminases. Four had received ICI as first-line therapy (Table 1). For patients 1, 2, and 5, a high PD-L1 tumor proportion score (>50%) returned earlier than ALK/ROS1/MET testing, prompting ICI initiation (Table 1). Patient 4 had recurrence of NSCLC shortly after completing adjuvant chemotherapy, and began nivolumab before MET testing resulted.

Table 1.

Detailed characteristics of patients who received crizotinib following ICI (n = 11).

ID	ICI	TKI	Oncogenic driver	ICI line	No. ICI doses	ICI-CRZ interval (days)	Liver met	PD-L1 TPS	PD-L1 Ab	ALT/AST elevation	TKI-liver tox interval (days)	Liver tox duration (days)	TKI interruption
1	Pembrolizumab + chemotherapy	Crizotinib	ALK	1	3	23	No	95%	22C3	Gr 4 ALT Gr 4 AST	38	37	Drug discontinued
2	Pembrolizumab + chemotherapy	Crizotinib	ALK	1	1	21	No	70%	22C3	Gr 4 ALT Gr 4 AST	27	44	Drug discontinued
6	Pembrolizumab	Crizotinib	ALK	1	1	10	No	60%	E1L3N	No	N/A	N/A	N/A
7	Pembrolizumab	Crizotinib	ROS1	2	3	27	No	20%	E1L3N	No	N/A	N/A	N/A
8	Nivolumab + ipilimumab	Crizotinib	ROS1	1	2	121	No	<1%	E1L3N	No	N/A	N/A	N/A
9	Atezolizumab + investigational drug	Crizotinib	ROS1	3	2	25	Yes	N/A	N/A	No	N/A	N/A	N/A
10	Nivolumab	Crizotinib	MET amp	2	12	73	No	0%	E1L3N	Gr 1 ALT Gr 1 AST	27	140	Dosing not interrupted
3	Nivolumab	Crizotinib	MET amp	2	4	30	Yes	50%	E1L3N	Gr 4 ALT Gr 4 AST	42	98	Drug discontinued
4*	Nivolumab	Capmatinib→Crizotinib + ibrutinib	MET ex 14	1	9	135	No	0%	E1L3N	Gr 3 ALT Gr 3 AST	49	61	Drug discontinued
5	Pembrolizumab	Crizotinib	MET ex 14	1	7	42	No	>90%	E1L3N	Gr 3 ALT Gr 2 AST	13	15	Held for 36 days; resumed with dose reduction
11	Pembrolizumab	Crizotinib	MET ex 14	3	2	99	No	95%	E1L3N	No	N/A	N/A	N/A

^*Patient received nivolumab followed initially by capmatinib. Soon thereafter, the patient was diagnosed with chronic lymphocytic leukemia (CLL) requiring discontinuation of capmatinib, and went onto receive crizotinib (for lung cancer) combined with ibrutinib (for CLL). The ALT/AST elevation was attributable to crizotinib (improved with holding crizotinib and recurred on re-challenge).

Abbreviations: ID, identifier; ICI, immune checkpoint inhibitor, TKI, tyrosine kinase inhibitor; ICI line, the line of systemic therapy for ICI; No. ICI doses, number of ICI doses the patient received; ICI-CRZ interval, the time interval between the last dose of ICI and start of crizotinib; liver met, presence of liver metastases prior to starting TKI; PD-L1 TPS, PD-L1 tumor proportion score; Ab, antibody; ALT, alanine aminotransferase; AST, aspartate aminotransferase; TKI-liver tox interval, the interval between the start of TKI and onset of liver toxicity; Gr, grade per the Common Terminology Criteria for Adverse Events version 4.0; ALK, anaplastic lymphoma kinase rearrangement; MET amp, MET amplification; MET ex 14, MET exon 14 skipping; ROS1, ROS1 rearrangement; N/A, not applicable.

The duration of hepatotoxicity in these five patients ranged between 15 and 98 days (median, 44 days). Four had concurrent grade 1 bilirubin elevation, and three had grade 3 alkaline phosphatase elevation (Table 1). No case met Hy’s law criteria. Four patients required permanent crizotinib discontinuation; one was able to resume crizotinib with dose reduction after holding for five weeks. All cases were reversible. None required steroids, and liver biopsies were not performed. There was no concomitant pneumonitis.

DISCUSSION

PD-1/PD-L1 inhibitors have become viable treatment options in addition to genotype-directed TKIs in oncogene-driven NSCLC. As a growing number of ICIs gain approvals in NSCLC, patients are increasingly likely to receive ICI therapy during their disease course. While the toxicity profiles of TKIs are well established, little is known regarding the potential toxicities of TKI administered after ICI.

In this study, we observed a higher incidence of hepatotoxicity with sequential ICI/crizotinib compared to crizotinib alone. The actuarial rates of grade 3/4 ALT and AST elevation with sequential therapy were 45.5% and 36.4%, respectively, versus 8.1% and 3.4% with crizotinib alone (consistent with the 3 to 10% reported in trials^11-13). Therefore, prior exposure to an ICI may significantly increase the risk of crizotinib-associated hepatotoxicity. While increased toxicities including transaminitis were observed in early-phase studies evaluating concurrent nivolumab plus crizotinib (CheckMate 370)⁶ and avelumab plus crizotinib (JAVELIN Lung 101),¹⁴ this is the first study to our knowledge to evaluate the effect of sequential ICI and TKI therapy on the risk of hepatotoxicity.

Our findings suggest that patients with ROS1/MET/ALK-driven NSCLC who receive crizotinib following prior ICI need to be cautioned and closely monitored for the development of hepatotoxicity. Of note, alectinib rather than crizotinib is now the standard first-line therapy in ALK-rearranged NSCLC, and the clinical impact of sequential ICI/alectinib was not evaluated in this study. It is worth noting, however, that in a phase Ib trial of concurrent alectinib and atezolizumab in ALK-rearranged NSCLC, an acceptable safety profile was observed, with grade 3 ALT increase occurring in 2 of 21 (9.5%) patients.¹⁵ These findings are in line with the acceptable tolerability of concurrent lorlatinib plus avelumab,¹⁴ yet are in contrast to the increased hepatotoxicities of crizotinib plus avelumab,¹⁴ crizotinib plus nivolumab,⁶ and ceritinib plus nivolumab.⁸ Indeed, the tolerability of different sequential or concurrent ICI/TKI strategies may be TKI-specific, urging further studies to elucidate which ICI/TKI combinations or sequences may be safe and viable options for patients. As suggested by the elevated hepatotoxicity seen with ICI plus crizotinib or ceritinib—versus alectinib or lorlatinib—the baseline level of hepatotoxicity seen with each TKI may impact the development of increased hepatotoxicity when used in a concurrent or sequential approach with ICI.

Our study had several noteworthy limitations. This was a retrospective study, subject to inherent selection biases and potential confounding by unmeasured risk factors. Further, the number of patients who received sequential ICI/crizotinib was small. Of note, although a higher proportion of the sequential ICI/crizotinib cohort had MET alterations and were smokers (possibly reflecting the association with MET alterations¹⁰) compared to the crizotinib-only cohort (Supplemental Table 1), the rate of hepatotoxicity is not known to be different in patients with MET-altered tumors.¹⁰ Additionally, it was not feasible to evaluate all sequences of available ICIs and TKIs; as mentioned above, it is likely that toxicities arising from sequential or concurrent ICI/TKI strategies may be TKI-specific. Finally, the mechanism by which prior ICIs may augment the risk of TKI-associated hepatotoxicity remains poorly understood. Further studies will help determine whether prior ICI may “prime” the immune system towards a TKI-induced inflammatory response.

In conclusion, prior immunotherapy may significantly increase the risk of crizotinib-associated hepatotoxicity. Careful monitoring for hepatotoxicity is therefore recommended in patients treated with sequential ICI and crizotinib. In parallel, further studies are needed to define the safety profile of different sequences of ICIs and TKIs and identify feasible treatment strategies in oncogene-driven NSCLC. Until a more fine-tuned understanding of specific ICI/TKI sequences is available, caution may be advised when clinicians face the real-world question of whether to utilize immunotherapy in patients with oncogene-driven NSCLC who may still have TKI options. In patients newly diagnosed with advanced-stage NSCLC, baseline comprehensive genotyping to identify targetable driver oncogenes should be expedited, and corresponding results incorporated along with the PD-L1 status in order to optimally inform treatment stratification and minimize potentially preventable toxicities.