1. FDA Evaluation of Hepatotoxicity Related to Tyrosine Kinase Inhibitors 1 Division of Drug Oncology Products; 2 Office of Surveillance and Epidemiology; 3 Office of Clinical Pharmacology; 4 Office of Oncology Drug Products J Chang, 1 M Rand, 2 G Blumenthal, 1 P Cortazar, 1 C Kulick, 2 E Hausman, 2 S Chang, 2 R Pratt, 2 B Habtemariam, 3 Y Chen, 3 J Bullock, 3 RC Orbach, 3 I Zineh, 3 R Justice, 1 R Pazdur 4

2. Background Tyrosine kinase inhibitors (TKIs) interfere with several critical signal transduction pathways in cancer As of March 2010, FDA had approved 9 TKIs (imatinib, dasatinib, nilotinib, erlotinib, gefitinib, lapatinib, sorafenib, sunitinib, and pazopanib) to treat a variety of malignancies All TKIs are associated with severe liver injury (SLI) and have different categories for liver injury, ranging from boxed warnings for fatal hepatotoxicity to adverse reactions for elevated LFTs LFTs=liver function tests. Adapted from Chang J et al. Presented at: ASCO Annual Meeting; June 3-7, 2011; Chicago, IL.

3. Methods  Analyses included a literature review and an evaluation of postmarketing reports of SLI associated with TKIs from FDA’s Adverse Event Reporting System* (AERS) from various dates of marketing approval until March 15, 2010. SLI is defined as acute liver injury (elevated LFTs, bilirubinemia, or jaundice) in combination with 1 of the following events: death, liver transplantation or placement on liver transplant list, hepatic encepholopathy, coagulopathy, or renal impairment  Drug utilization data were reviewed for comparison across TKIs. The estimated number of nationally projected dispensed prescriptions from mail order/specialty pharmacy and outpatient retail pharmacy settings was obtained from Wolters Kluwer Source PHAST Prescription Monthly™ from 2006 through 2009  For a mechanistic explanation, laboratory and exposure data were reviewed from 8 new drug application (NDA) submissions. Gefitinib data are excluded because it is no longer marketed in the United States. Exposure-response properties of TKI-induced liver enzyme and bilirubin elevations were evaluated *Limitations of AERS data include: underreporting, stimulated reporting, biased reporting, variable quality and content, and no certainty that the reported event is due to the product. Adapted from Chang J et al. Presented at: ASCO Annual Meeting; June 3-7, 2011; Chicago, IL.

4. Results Literature search: As of April 2010, there were 19 case reports of liver injury with imatinib, 7 with erlotinib, 5 with sorafenib, 4 with sunitinib, and 2 with gefitinib. 15 of these reports were captured in AERS Drug utilization results: Total oral TKI market increased from approximately 313,000 prescriptions in 2006 to 348,000 prescriptions in 2009. Imatinib had the largest proportion of dispensed prescriptions Adapted from Chang J et al. Presented at: ASCO Annual Meeting; June 3-7, 2011; Chicago, IL.

5. Results (cont’d) TKIFDA Approval Year Total Prescriptions 2006-2009, N Market Share 2006-2009, % Imatinib2001652,99745.1 Erlotinib2004479,11433.1 Sunitinib2006148,86410.3 Lapatinib200766,0084.6 Dasatinib200647,3533.3 Gefitinib200317,3251.2 Sorafenib200525,7261.8 Nilotinib20079,6710.7 Pazopanib20091050 TOTAL1,447,163100 Adapted from Chang J et al. Presented at: ASCO Annual Meeting; June 3-7, 2011; Chicago, IL. Source: Wolters Kluwer SOURCE PHAST Prescription, 2005-2009. Total Number of Nationally Projected Dispensed Prescriptions for TKIs in US Mail-Order and Outpatient Retail Pharmacies, 2006-2009

6. Postmarketing Cases Of the 497 cases evaluated for SLI, 55 were categorized as possibly or likely associated with a drug The majority of SLI cases were associated with sunitinib (22) and imatinib (13) Most common event terms were hepatic failure, hepatic necrosis, hepatic coma or encephalopathy, and hepatorenal failure 67% of postmarketing cases were from foreign sources, and 45% of cases involved patients enrolled in clinical trials No dasatinib cases were classified as possibly or likely associated with a drug Adapted from Chang J et al. Presented at: ASCO Annual Meeting; June 3-7, 2011; Chicago, IL.

7. Postmarketing TKI Cases of SLI* * Limitations of AERS data include: underreporting, stimulated reporting, biased reporting, variable quality and content, and no certainty that the reported event is due to the product. † Gefitinib is no longer marketed in the United States. CML=chronic myelogenous leukemia; GIST=gastrointestinal stromal tumor; NSCLC=non–small cell lung cancer; RCC=renal cell carcinoma. Adapted from Chang J et al. Presented at: ASCO Annual Meeting; June 3-7, 2011; Chicago, IL. Sunitinib n=22 Imatinib n=13 Gefitinib † n=7 Erlotinib n=6 Sorafenib n=2 Nilotinib n=2 Lapatinib n=1 Pazopanib n=2 Demographics Age (years) Median Range 64 45-83 50 4-78 72 48-88 67 53-7756 38 22-5354 29 21-37 Gender Male Female Unknown 14 8 1 10 3 4 331 2020 0101 0202 Cancer type RCC 16 GIST 2 Other 4 CML 8 GIST 3 Other 2 NSCLC 6 Pancreatic 1 NSCLC 5 Other 1 Thyroid 1 RCC 1 CML 1 GIST 1 Breast 1 Synovial sarcoma 2

8. Postmarketing TKI Cases of SLI* (cont’d) *Limitations of AERS data include: underreporting, stimulated reporting, biased reporting, variable quality and content, and no certainty that the reported event is due to the product. † Gefitinib is no longer marketed in the United States. ‡ 2 Patients required liver transplant. § Other includes: hepatitis fulminant (2), liver disorder (1), hepatitis (1), hepatic cirrhosis (1). Adapted from Chang J et al. Presented at: ASCO Annual Meeting; June 3-7, 2011; Chicago, IL. Sunitinib n=22 Imatinib n=13 Gefitinib † n=7 Erlotinib n=6 Sorafenib n=2 Nilotinib n=2 Lapatinib n=1 Pazopanib n=2 Reported Adverse Event and Event Characteristics Hepatic failure Hepatic necrosis Hepatic coma or encephalopathy Hepatorenal failure Other § 16 2 2 0 2 10 ‡ 1 0 1 1 4010240102 5001050010 1010010100 2000020000 1000010000 2000020000 Dose (mg) Median Range 50 25-50 400 200-400 250 None 150 100-150 800 None 1500 None 800 None Time to event (days) Median Range 27 7-314 150 8-312 22 4-180 28 12-71 68 56-80 63 6-120126 45 34-55 Outcome Death Hospitalization Life-threatening Other 16 4 2 0 85088508 61006100 51005100 10011001 20002000 01000100 20002000 (+) dechallenge (+) rechallenge 5050 1010 0202 1010 10100 10100

9. Pharmacogenomics HLA=human leukocyte antigen. Adapted from Chang J et al. Presented at: ASCO Annual Meeting; June 3-7, 2011; Chicago, IL. This is being explored as a potential tool to understand the mechanistic basis of drug-induced liver injury (DILI) and manage its clinical risks Susceptibility to idiosycratic DILI is currently attributed to the complex interaction of multiple factors such as genetic, environmental, and drug-related factors Genetic susceptibility to DILI is mostly associated with variations in immune response and/or drug metabolism and transport genes Lapatinib-induced hepatotoxicity has recently been linked to a HLA allele

10. Examples of Genotypic Associations for DILI DrugTherapeutic ClassHLA AlleleReference (PMID) FlucloxacillinAntibioticHLA-B*5701 Daly et al. 2009 [19483685] XimelagatranThrombin inhibitor HLA-DRB1*07 and HLA-DQA1*02 Kindmark et al. 2008 [17505501] LumiracoxibNSAIDHLA-DQA1*0102 Singer et al. 2010 [20639878] LapatinibAntineoplasticHLA-DQA1*0201 Spraggs et al. 2011 [21245432] NSAID=nonsteroidal anti-inflammatory drug; PMID=PubMed identifier. Adapted from Chang J et al. Presented at: ASCO Annual Meeting; June 3-7, 2011; Chicago, IL.

11. Clinical Pharmacology Adapted from Chang J et al. Presented at: ASCO Annual Meeting; June 3-7, 2011; Chicago, IL.  Preliminary review of the data from NDA submissions suggests that TKIs inhibit bilirubin clearance. As shown below, sorafenib (sor), nilotinib (Nil), and pazopanib (pazo) exhibit concentration-dependent inhibition of the hepatic UGT1A1 enzyme in vitro. Inhibition of UGT1A1 is expected to increase total bilirubin in vivo. % Inhibition 0 100 0.51.010.05.0 Drug Concentration (  M) 80 40 60 20 sor:estradiol sor:SN-38 Nil:bilirubin Nil:estradiol pazo:HFC 0.150.0100.0

12. Clinical Pharmacology (cont’d) ALT=alanine aminotransferese; CI=confidence interval; ULN=upper limit of normal. Adapted from Chang J et al. Presented at: ASCO Annual Meeting; June 3-7, 2011; Chicago, IL. With increasing trough concentrations of pazopanib, the probability of ALT >3 x ULN increases with increasing pazopanib plasma concentrations No concentration-dependent increase in total bilirubin >2 x ULN is observed Probability of ALT >3 x ULN With Increasing Pazopanib Concentrations Logistic regression Observed proportion (95% CI) Probability of bilirubin >2 x ULN With Increasing Pazopanib Concentrations Probability of ALT >3 x ULN (%) 0 40 60 80 100 02060 20 8040 0 60 80 02060 20 8040 Probability of ALT >3 x ULN (%) Steady State Trough Concentration (  g/mL) Logistic regression Observed proportion (95% CI) 16/63 17/63 11/63 8/64 1/64 3/63 2/63 4/63

13. Conclusions *Limitations of AERS data include: underreporting, stimulated reporting, biased reporting, variable quality and content, and no certainty that the reported event is due to the product. Adapted from Chang J et al. Presented at: ASCO Annual Meeting; June 3-7, 2011; Chicago, IL.  The pharmacogenomics data for lapatinib thus far suggest a genetic association for DILI. More research into the predictive value, sensitivity, and specificity of these biomarkers is needed to understand their clinical utility for predicting patients who are susceptible  These associations may be used in the future to identify individuals at greater risk of developing liver injury  Consideration of both genetic and nongenetic factors in DILI may help to identify these patients and improve therapy management. Investigations of the exposure response analysis for other TKIs are ongoing  Review of postmarketing and clinical pharmacology data suggests a role for TKIs in causing hepatotoxicity. However, review of postmarketing AERS data* precludes class labeling for SLI for all TKIs  Greater numbers of cases for imatinib and sunitinib may relate to greater national market share of TKIs

14. TKI Labeling for Hepatotoxicity *Imatinib and sorafenib prescribing information updated in italics. † Gefitinib is no longer marketed in the United States. Adapted from Chang J et al. Presented at: ASCO Annual Meeting; June 3-7, 2011; Chicago, IL. Drug Boxed Warning Warnings and Precautions Adverse Reactions Additional Data From Clinical Trials or Postmarketing Experience Imatinib* Fatal Hepatotoxicity, Severe Liver Injury (requiring liver transplant), Acute Liver Failure, Elevated LFTs Elevated LFTs Hepatitis, Hepatic Failure, and Hepatic Necrosis (including some fatalities) DasatinibElevated LFTsHepatitis NilotinibElevated LFTs Hepatitis, Hepatotoxicity Erlotinib Hepatic Failure and Hepatorenal Syndrome (including fatalities) Elevated LFTs Gefitinib † Elevated LFTs (Precautions section) LapatinibFatal Hepatotoxicity Hepatotoxicity SunitinibFatal Hepatotoxicity Hepatotoxicity Sorafenib*Liver Dysfunction Drug-induced hepatitis, including reports of hepatic failure and death PazopanibFatal Hepatotoxicity Elevated LFTs Updated changes are in bold and italics