The best magazine
Genetic Characterization of ALT Elevation During Pazopanib or Lapatinib Treatment
Pharmacogenetics has provided key mechanism insights for DILI with the discovery of genetic associations for drug metabolism genes and, more recently, HLA alleles for DILI caused by a diverse range of medicines, including the commonly used antibiotics flucloxacillin and amoxicillin–clavulanate. The HLA associations were not indicated by the pharmacological properties of the drugs and link DILI to metabolic transformation and immunological mechanisms. This is consistent with other reports of drug-induced hypersensitivity adverse reactions, such as abacavir-induced hypersensitivity and carbamazepine-induced Stevens–Johnson syndrome (SJS)–toxic epidermal necrosis (TEN).
Pazopanib-induced ALT elevation (>3× ULN) occurred in 17% of treated patients with RCC. In 93% of affected patients, ALT elevations occurred within 18 weeks of treatment. A two-stage pharmacogenetic study in RCC patients receiving pazopanib investigated genetic variation in 282 candidate genes, selected for involvement in pazopanib biological pathways, pazopanib metabolism and disposition and liver function. Following 'discovery' and 'confirmation' analysis, two correlated variants from HFE were found to be significantly associated with ALT elevation. The frequencies of the HFE rs707889 TT risk genotype are approximately 3% in Caucasians and Asians, and 2% in Hispanics. Figure 3 displays liver chemistry data as an 'evaluation of drug-induced serious hepatotoxicity' (eDISH) plot for pazopanib-treated patients to identify individual patients with liver chemistry elevations of possible concern. The eDISH plot shows patients with elevated ALT and/or TBL during pazopanib treatment, one of whom had concurrent ALT/TBL elevations (Figure 3). In addition, this eDISH plot has been annotated to show HFE rs707889 genotype carriage. As described above, there appears to be an enrichment of the TT genotype with higher ALT values, and it is noted that the patient with concurrent ALT/TBL elevations was not a carrier of the implicated HFE genotype. HFE encodes a membrane protein that regulates iron homeostasis. Genetic mutations in HFE, most notably the loss-of-function mutation C282Y, are associated with hereditary hemochromatosis, an iron storage disorder with increased absorption of dietary iron and excess tissue iron storage. Although none of the ALT elevation cases were homozygous for the C282Y mutation, an increased carriage of the C282Y heterozygous genotype was observed in ALT cases (33%) compared with controls (17%). Among ALT cases that carried the HFE rs707889 TT risk genotype (17% of ALT cases), recovery from ALT elevation occurred in 91% of these patients, with the remaining 9% of ALT cases having limited or no follow-up data to enable recovery determination. In addition, recovery was maintained in some ALT cases while continuing pazopanib treatment (adaptation) and following rechallenge after discontinuation (negative rechallenge). None of the patients carrying the HFE rs707889 TT risk genotype had concurrent ALT/TBL elevation, or developed liver failure. These observations are consistent with the hypothesis that pazopanib-induced ALT elevations that are associated with HFE variants may represent a change in liver function with moderate isolated ALT elevation that resolves (i.e., adaptation) and may not warrant discontinuation during cancer treatment. The present analysis did not identify HLA alleles as associated with pazopanib-induced ALT elevations, suggesting a different mechanism of action for pazopanib-associated transaminase elevation compared with DILI caused by other drugs, including flucloxacillin, amoxicillin–clavulanate and also lapatinib (see below).
(Enlarge Image)
Figure 3.
Evaluation of drug-induced hepatotoxicity for pazopanib-treated patients with renal cell carcinoma by HFE genotypes.
These data are pooled from Caucasian patients who participated in two advanced/metastatic renal cell carcinoma clinical trials and received pazopanib treatment (n = 236). Each point in the figure represents an individual patient's on-treatment maximum values for serum ALT (x-axis) and TBL (y-axis), as ×ULN. Each patient has been annotated by their HFE rs707889 genotype: G/G or G/T (blue open circle) and T/T (red filled circle). The T/T genotype is associated with increased ALT elevations during pazopanib treatment. Note that one patient (black circled) had concurrent ALT and TBL elevations consistent with classification as a possible Hy's law case; this case was not a carrier of the HFE T/T genotype.
×ULN: Multiplied by the upper limit of normal; TBL: Total bilirubin.
In the majority of lapatinib-associated liver safety events, regular liver chemistry monitoring and adherence to treatment discontinuation guidance results in their resolution. A retrospective case review demonstrated a high incidence of positive rechallenge (11 out of 20 cases; 55% [GlaxoSmithKline, Data on File]), with recurrent ALT elevations that meet established criteria, when lapatinib had been reintroduced following discontinuation and resolution of prior liver chemistry elevations. The slow rate of accumulation of ALT elevation cases following initiation of lapatinib treatment (median: 110 days), and the high rate of positive rechallenge, are potential hallmarks of immune-mediated liver injury. A comprehensive pharmacogenetic investigation of ALT elevation has been conducted in lapatinib-treated patients with metastatic breast cancer. In a two-stage case–control analysis with extensive genotyping to include HLA genes, 300 candidate genes selected for involvement in lapatinib pharmacology or metabolism and disposition and genes involved in liver function plus genome-wide screening of 1 milion common SNPs, were evaluated for differences in carriage frequency between subjects with treatment-associated ALT elevation in a 'discovery' data set of ALT cases and treated controls, selected from 12 lapatinib-treatment clinical trials of metastatic breast cancer patients. Three HLA variants in a single class II MHC locus, centered on HLA-DQA1*02:01 and -DRB1*07:01, were identified and remained associated with lapatinib-induced ALT elevation in an independent data set. This DQA1/DRB1 locus association is the same as previously reported for ximelagatran, a direct thrombin inhibitor that was terminated for serious hepatotoxicity during Phase III clinical development. No additional genetic signals comparable in strength to these HLA associations have been identified for lapatinib from analysis of the candidate genes or from genome-wide screening in the available data set. Notably, no associations were observed in the present data set for HFE variants implicated in pazopanib-induced ALT elevation, or for transporter gene variants implicated in cholestatic liver injury (BSEP/ABCB11,ABCB1, BCRP/ABCG2 or SLCO1B1). The HLA association was confirmed in an independent trial that compared treatment of lapatinib combined with the aromatase inhibitor letrozole versus letrozole alone. This produced a robust association of DQA1*02:01/DRB1*07:01 allele carriage with ALT elevation risk (odds ratio: 9) during lapatinib treatment. HLA class II peptides form heterodimer proteins for which DQA1/DQB1 and DRA/DRB1 combinations create discrete antigen binding sites.HLA-DQA1*02:01/DRB1*07:01 are highly correlated and further investigation has not dissected which allele is causative for lapatinib-associated hepatotoxicity. The DQA1*02:01/DRB1*07:01 association was not observed for ALT elevation observed in the letrozole-only treatment group, consistent with a specific association for lapatinib. Furthermore, differences in the cumulative incidence of ALT elevation (≥3× ULN) between DQA1*02:01/DRB1*07:01 allele carriers and noncarriers in the lapatinib-containing treatment regimen suggested that removal of the specified HLA allele carrier patients from this treatment would reduce the rate of ALT elevation by more than two-thirds, to a level comparable to that observed for letrozole treatment alone.
While drug-induced ALT elevations to >3× ULN are recognized as important signals of liver toxicity and potential injury, benefit/risk considerations in the treatment of metastatic breast cancer do not result in treatment discontinuation for such signals. Serum liver chemistry levels in patients receiving lapatinib-treatment are monitored regularly and treatment is discontinued for isolated ALT elevations that achieve >5× and >8× ULN for adjuvant and metastatic breast cancer, respectively, or concomitant ALT (>3× ULN) and TBL (>2× ULN) in both disease settings. Therefore, we sought to evaluate the association of the specified HLA alleles with more serious liver safety signals that result in discontinuation and withdrawal of lapatinib treatment. The specified HLA allele association was maintained and strengthened for higher ALT elevation thresholds. Review of clinical trial data from approximately 10,000 patients evaluating lapatinib treatment regimens in adjuvant and metastatic breast cancer identified 20 patients with potentially serious lapatinib-related liver injury (e.g., probable/possible Hy's law cases) and for whom DNA was available. Figure 4 shows a modified eDISH plot that displays only these 20 cases with concurrent ALT and bilirubin elevations. These cases showed a strong association for the previously implicated class II HLA alleles with an allele carriage frequency of 70%. Stratification by carriage/noncarriage of the specified HLA alleles suggested differences in the magnitude of liver safety signal (isolated ALT elevation) and other characteristics (Figure 4). All cases that attained grade 3 ALT elevation (>5× ULN; n = 11) were carriers of the specified HLA alleles. For the group of HLA noncarriers, the majority (five out of six) had underlying Gilbert's syndrome (based on UGT1A1*28/*28 homozygous genotype), which likely contributed to their concurrent TBL elevation. In some of these patients, TBL elevation was observed to precede their corresponding ALT elevation, which is inconsistent with initial serious hepatocyte injury resulting in impaired bilirubin excretion. A high incidence of Gilbert's syndrome was anticipated in these cases, since serious liver injury is defined by concomitant ALT and TBL elevations, and Gilbert's syndrome-related TBL elevations may be classified erroneously as potential serious liver injury (or Hy's law) if accompanied by modest ALT elevations, as indicated in these specified HLA-negative cases. These data support association of the specified HLA alleles with cases of serious hepatotoxicity that resulted in their permanent discontinuation of lapatinib treatment. Furthermore, information derived from specified HLA and UGT1A1*28 genotyping emphasized the importance of the Gilbert's syndrome as a potential confounder in the diagnosis of serious liver injury (or Hy's law), reinforcing that serum bilirubin fractionation or, if not available, the UGT1A1*28/*28 genotype, should be determined to enable the fullest characterization in all cases of drug-induced concomitant ALT/TBL elevation.
(Enlarge Image)
Figure 4.
Cases of lapatinib-associated serious liver injury.
Evaluation of drug-induced serious hepatotoxicity plot of concurrent ALT/TBL cases during lapatinib treatment, classified by the presence of specified HLA alleles and UGT1A1*28/*28 genotypes. Evaluation of 20 cases of concurrent ALT/TBL elevation, adjudicated as 'possible/probable' drug-induced liver injury. Each point represents an individual case patient's maximum concurrent serum ALT and TBL values (as ×ULN), shown on the x- and y-axes, respectively. In addition, each point is coded for the presence of the specified HLA alleles (positive or negative for DQA1*02:01/DRB1*07:01) and UGT1A1*28 homozygous genotype (positive or negative for *28/*28 genotypes, which is associated with Gilbert's syndrome).
×ULN: Multiplied by the upper limit of normal; TBL: Total bilirubin.
Reproduced with permission from [93].
The class II HLA locus association with lapatinib-induced hepatotoxicity suggests an immune-mediated hepatocyte injury, involving CD4-positive T cells. It is speculated that a lapatinib-derived reactive metabolic product(s) binds to a liver protein(s) to form an antigenic hapten(s). Possible candidates for hapten formation may be the liver CYP3A4 and CYP3A5 isoenzymes that are responsible for the majority of phase I metabolism of lapatinib. Studies have demonstrated mechanism-based inactivation of CYP3A4 and CYP3A5 by lapatinib, where highly reactive quinone-imine or amine metabolic products bind pseudoirreversibly to the CYP3A isoforms, and may render them susceptible to antigen processing, hapten formation and immune recognition. It is hypothesized that the hapten is recognized by antigen-presenting cells in the liver in individuals that have restricted expression of HLA-DQA1*02:01 and/or -DRB1*07:01 class II heterodimers. This recognition results in downstream activation of CD4-positive T cells through presentation of the hapten to the T-cell receptor complex and the development of a targeted immune response within the liver. There is general support for this mechanism from ex vivo studies demonstrating class I HLA-restricted CD8-positive T-cell activation as the immunogenetic basis of drug-induced adverse events, including abacavir-induced multiorgan hypersensitivity and carbamazepine-induced SJS/TEN. To date, similar experiments have not demonstrated a class II HLA-restricted, CD4-positive, T-cell activation on drug challenge ex vivo. The observation that a common, specified HLA association is necessary but not sufficient to elicit the rare adverse event of lapatinib-induced serious liver injury, or conversely, that the majority of HLA-DQA1*02:01/DRB1*07:01-allele-bearing patients treated with lapatinib do not experience liver injury, suggests other factors are involved to restrict this toxicity to a small subset of the specified HLA allele carriers. Such factors may include: restricted production of a reactive metabolite, restricted hapten formation and presentation, tightly controlled immune regulation with dominant self-tolerance occuring in the majority of HLA-positive patients; and activation of repair mechanisms. Additional polymorphic components may be present in the subset of HLA-positive, at-risk patients. For example, in carbamazepine-induced SJS/TEN adverse events, the hapten-mediated immune reaction has a requirement for the correct combination of restricted TCR, present on CD8-positive T cells and restricted class I HLA alleles on antigen-presenting cells for pathology to occur. Isolation of activated T-cell subsets from lapatinib-sensitized patients would enable next-generation, deep sequencing of the highly polymorphic TCR and identification of restricted TCR repertoires that may further define this adverse event mechanism. If used in combination with the specified HLA marker, TCR testing might increase predictive sensitivity of a risk biomarker for lapatinib hepatotoxicity.
The robust, confirmed association of DQA1*02:01/DRB1*07:01 alleles with lapatinib-induced isolated ALT elevation and serious hepatocellular injury with hyperbilirubinemia suggests specified HLA testing might have some opportunity for clinical utility in reducing the incidence of lapatinib-induced hepatotoxicity by identifying patients at increased risk and informing their clinical management options in the challenging disease setting of breast cancer. Allele carriage of DQA1*02:01/DRB1*07:01 is relatively common in major ethnic groups, with frequency ranging from 25 to 15% in white, African–American, Hispanic and Asian patients. It is noted that the same database reported low frequency of the DQA1*02:01/DRB1*07:01 allele carriage in subjects of Japanese descent, however, limited exposure of lapatinib in Japanese cancer patients has not enabled assessment of the impact of low carriage frequency of the specified HLA alleles. Although the specified HLA alleles demonstrate a suitably high negative predictive value (NPV ≥0.97) for hepatotoxicity risk, a limitation to prospective HLA testing in lapatinib-treated patients is the relatively low positive predictive value (≤0.17), translating to a high false-positive rate, where HLA testing might lead to exclusion of metastatic breast cancer patients from lapatinib treatment, many of whom who might benefit and suffer no harm. Therefore, prospective, specified HLA testing is not recommended for the approved indications of lapatinib in the treatment of metastatic breast cancer. This challenge to the clinical application of HLA testing to predict risk of serious drug-induced adverse events has been noted previously. Nevertheless, the high NPV translates to considerably lower hepatotoxicity risk in noncarriers of the specified HLA alleles, and does suggest some clinical utility for the HLA test to assess causality of DILI in lapatanib-treated patients receiving other treatments, possibly by discriminating lapatinib from other liver injury causes, including other commonly used drugs (e.g., flucloxacillin and amoxicillin–clavulanate) and concurrent chemotherapy (e.g., capecitabine and taxanes). Where no additional drug is considered causative, liver chemistry changes in lapatanib-treated patients who are negative for the specified HLA alleles may suggest a nonimmune-related cause for liver chemistry abnormalities and may enable consideration of reintroduction of lapatinib in such patients to maintain disease treatment. In addition, specified HLA testing might be used to identify a subgroup of patients with higher risk for lapatinib hepatotoxicity, who could be scheduled for more frequent liver chemistry monitoring and early liver safety signal detection. This would focus attention on patients with the highest risk of lapatinib-induced hepatotoxicity and also reduce the monitoring burden for the majority (>75%) of patients who are negative for the specified HLA alleles. With frequent monitoring in specified HLA allele-positive patients, a combination of an early liver safety signal (e.g., ALT >3× ULN) plus carriage of the specified HLA allele may suggest earlier treatment discontinuation to avert subsequent, serious liver injury. Further work is ongoing to validate and assess the predictive utility of this potential liver safety biomarker in a prospectively defined, retrospectively conducted analysis in a large, randomized clinical trial evaluating lapatinib monotherapy versus placebo as adjuvant treatment in early-stage breast cancer. In addition, whole-genome deep sequencing of previously identified lapatinib-induced possible Hy's law cases is underway to attempt to identify additional genetic determinants for hepatotoxicity risk.
Source: ...