1 Introduction
Autosomal dominant polycystic kidney disease (ADPKD) is an inherited disorder characterized by the appearance and slow growth of fluid-filled cysts primarily in the kidneys, but also in the liver and other organs [
1,
2]. Over time, the expanding cysts physically displace and obstruct renal tubules, blood vessels and lymphatics, as well as promote apoptosis, atrophy and fibrosis of the renal parenchyma, leading to progressive renal failure [
3]. ADPKD was responsible for 2.6 % of patients on dialysis and 9.9 % of patients receiving renal transplants in the USA in 2012 [
4].
Studies of animal models implicate the antidiuretic hormone arginine vasopressin and its secondary messenger 3′,5′-cyclic adenosine monophosphate (cAMP) as promoters of kidney-cyst cell proliferation and luminal fluid secretion [
5,
6]. In early animal models the suppression of vasopressin release by vasopressin V
2-receptor inhibition slowed disease progression [
7,
8]. Because tolvaptan is a V
2 receptor antagonist, its use in treating ADPKD was investigated. In the pivotal TEMPO 3:4 (Tolvaptan Efficacy and Safety in Management of Autosomal Dominant Polycystic Kidney Disease and its Outcomes) trial (ClinicalTrials.gov identifier: NCT00428948 [
24]), long-term treatment with tolvaptan was associated with favorable outcomes in subjects with early ADPKD. These favorable outcomes included lower rate of growth in total kidney volume (2.8 vs. 5.5 %;
p < 0.001) and slower decline in kidney function [reciprocal of the serum creatinine level, −2.61 vs. −3.81 (mg per mL)
−1 per year;
p < 0.001] [
9].
TEMPO 3:4 and its extension trial TEMPO 4:4 (ClinicalTrials.gov identifier: NCT01214421 [
25]) were continuously monitored by an Independent Data Monitoring Committee (IDMC) that assessed overall safety, including hepatic function and injury. During the course of continued monitoring of the TEMPO trials, the IDMC recommended increasing the frequency of liver monitoring in the TEMPO 4:4 extension study, from every 6 months to every 3 months. Upon unblinding of the TEMPO 3:4 database, an imbalance in hepatic injury, defined as alanine aminotransferase (ALT) >3 times the upper limit of normal (ULN), was observed for subjects receiving tolvaptan (4.4 %) relative to placebo (1.0 %). To assess the risk of hepatotoxicity, an independent, blinded, expert Hepatic Adjudication Committee (HAC) re-examined subject-level data from all ADPKD clinical trials, as well as data from non-ADPKD subjects who had received long-term tolvaptan therapy in other clinical trials, including patients with hyponatremia [syndrome of inappropriate anti-diuretic hormone release (SIADH)], heart failure, and cirrhosis. The results of this analysis are presented here.
2 Methods
2.1 Subjects
The safety databases reviewed here were generated in clinical trials that examined the efficacy and safety of tolvaptan in the treatment of ADPKD. The vast majority of included subjects were from the pivotal, randomized, placebo-controlled TEMPO 3:4 trial (NCT00428948 [
24]) [
9] and its open-label extension TEMPO 4:4 (NCT01214421 [
25]). To be eligible for inclusion in TEMPO 3:4, all subjects had to have an image-confirmed diagnosis of ADPKD, total kidney volume ≥750 mL and an estimated creatinine clearance rate ≥60 mL/min. 1445 subjects were enrolled (tolvaptan, 961; placebo, 484) and 1441 had at least one post-baseline assessment of hepatic injury. Tolvaptan was administered twice daily, starting at a morning/afternoon dose of 45/15 mg and titrated weekly to 60/30 mg and 90/30 mg based on tolerability; subjects could down-titrate at any time to as low as the starting dose (45/15 mg). Subjects were treated for 36 months or until early discontinuation. In TEMPO 4:4, 871 subjects from the lead TEMPO 3:4 trials (tolvaptan, 557; placebo, 314), received open-label tolvaptan at their highest tolerated dose for a minimum of 24 additional months. The study drug was discontinued when pre-determined criteria were met. Re-challenge was conducted on an individual basis.
To provide further context for this analysis, safety databases from the non-ADPKD clinical trials evaluating the efficacy and safety of tolvaptan in subjects with heart failure or hyponatremia (etiologies included SIADH, heart failure, and cirrhosis) were also examined. In total, 6155 subjects were enrolled (tolvaptan, 3403; placebo, 2752) across all non-ADPKD trials, and 4664 subjects (tolvaptan, 2414; placebo, 2250) were enrolled in long-term placebo-controlled trials. All trials were sponsored by Otsuka Pharmaceuticals.
2.2 Evaluation of Drug-Induced Serious Hepatotoxicity (eDISH) Assessments
The Evaluation of Drug-Induced Serious Hepatotoxicity (eDISH) approach was used to efficiently visualize potential hepatocellular injury in the large subject populations examined in this study and to supply supportive data for the adjudication process. This is a graphical methodology in which the log of the peak serum ALT concentration is plotted for each subject along the x-axis and the log of peak total serum total bilirubin (BT) concentration is plotted along the y-axis [
10]. Four quadrants on the eDISH plot are defined by lines at 3 × ULN for ALT and 2 × ULN for BT. The upper-right quadrant is the Hy’s Law quadrant, although patients may also appear there due to cholestatic liver injury. To separate out these latter confounders, US FDA guidance defines a subject in the upper-right quadrant as having severe drug-induced liver injury (DILI) when the serum alkaline phosphatase is <2 × ULN [
11].
An excess of subjects in the lower-right quadrant for a study drug relative to placebo also indicates a drug that may be capable of causing liver injury, even when examination of the Hy’s Law quadrant is unrevealing [
11]. This reflects the fact that ALT is a more sensitive indicator of hepatocellular injury than BT and that increases in ALT may occur before or without accompanying rises in serum bilirubin [
12]. Subjects discovered to have elevated serum ALT may be discontinued from treatment before the injury progresses to the point of bilirubin elevations. Finally, an excess of patients in the upper-left quadrant for a study drug relative to placebo is generally observed in drugs associated with cholestatic liver reactions, or patients with Gilbert’s syndrome or other causes of isolated hyperbilirubinemia.
2.3 Adjudication of Hepatic Safety Signals
The HAC consisted of four expert hepatologists (PBW, JHL, NK, and DHA) who reviewed safety data from ADPKD and non-ADPKD (hyponatremia, heart failure, and cirrhosis) tolvaptan trials. Five general categories of adverse events were identified for further adjudication:
1.
Category 1 A non-serious treatment-emergent event leading to discontinuation of treatment or any serious treatment-emergent adverse event matching a lower level term in any one of the following hepatic standard MedDRA
®
1 queries (SMQs):
(a)
cholestasis and jaundice of hepatic origin,
(b)
hepatic failure, fibrosis, and cirrhosis and other liver damage-related conditions,
(c)
non-infectious hepatitis,
(d)
liver-related investigations, signs, and symptoms,
(e)
liver-related coagulation and bleeding disturbances.
2.
Category 2 ALT >3 × ULN and BT >2 × ULN.
3.
Category 3 aspartate aminotransferase (AST) >3 × ULN and BT >2 × ULN.
4.
Category 4 ALT or AST >5 × ULN.
In cases where ULN could not be obtained, 40 IU/L and 1 mg/dL (17 µM/L) were used for ALT and BT, respectively. Laboratory evaluations were completed during the TEMPO 3:4 trial at baseline, 3 weeks, 4 months, and every 4 months thereafter.
For causality assessment, the four hepatologists comprising the HAC agreed to utilize “expert opinion” rather than a structured scoring instrument [e.g., Roussel Uclaf Causality Assessment Method (RUCAM)] [
13]. The committee assessed causality of all adjudicated events based on co-morbid conditions, concomitant medication use, onset, offset, and dose relationship. Events of interest were allocated into the following five causality groups, as defined by the Drug-Induced Liver Injury Network [
14,
15]: “definite”, i.e., evidence that the drug is causing the injury is beyond reasonable doubt; “highly likely”, i.e., evidence that the drug is causing the injury is clear and convincing but not definite; “probable”, i.e., the preponderance of evidence supports the link between the drug and liver injury; “possible”, i.e., the evidence for the drug causing the injury is equivocal but present; and “unlikely”, i.e., evidence is available that an etiological factor other than the drug caused the injury. Cases judged “not classifiable” in regard to the role of the study drug occurred when the data were insufficient to render an opinion. Such situations arose when other confounding factors were present (e.g., diseases or other medications) but details could not be provided for the intensity/dose and timing of the underlying conditions (e.g., congestive heart failure, ischemia, multiple drug administration, underlying liver disease). Insufficient data were infrequently encountered [0/98 (0 %) ADPKD cases and 5/52 (10 %) non-ADPKD cases], but when it was the consensus opinion, it was always in cases with significant underlying diseases other than ADPKD.
The four hepatologists on the committee performed their assessments independently and without knowledge of treatment assignment. If committee members did not arrive at an identical independent causality score for a particular case, the case was discussed further in conference and a final causality determination was achieved by consensus, although majority rule was used in rare instances where consensus could not be achieved.
Because TEMPO 4:4 was ongoing at the time the adjudication was initiated, the committee performed two rounds of adjudication, one with a cut-off date of 31 March 2012 to provide initial feedback and guide future treatment regimens, and a later one with a cut-off date of 31 March 2015. The selection criteria for adjudicated cases were identical during the two periods, with the exception that Category 5 (BT >2 × ULN) was not employed between October 2012 and February 2014. This change was prompted by discussions among the hepatology experts, who agreed that a patient with an isolated elevated serum bilirubin in the absence of the other selection criteria was not a safety concern.
4 Discussion
No cases of acute liver failure have been reported in clinical trials of tolvaptan in ADPKD, and all subjects experiencing hepatic injury have recovered. Nonetheless, tolvaptan treatment was associated with elevations in serum aminotransferases exceeding 3 × ULN in subjects with ADPKD. This finding alone is not considered to be a reliable liver safety signal [
11], but three tolvaptan-treated subjects in the TEMPO 3:4 clinical trial and its open-label extension were confirmed to have met criteria for Hy’s Law (ALT >3 × ULN and BT >2 × ULN). Based on FDA guidance [
11], the identification of Hy’s Law cases indicates that tolvaptan has the potential to cause hepatic injury capable of progressing to liver failure in patients with ADPKD. The apparent rarity of hepatic injury observed with tolvaptan has all the characteristics of an idiosyncratic reaction, suggesting the vast majority of patients should be able to receive long-term treatment without risk of liver injury.
A signature presentation of liver injury was discerned from the reviewed data. Tolvaptan-associated hepatocellular injury had an onset between 3 and 18 months, with little evidence of similar events before or after this period of apparent susceptibility. The injury typically progressed by biochemical criteria for a median of 28 [interquartile range (IQR) 15–50] days after discontinuation of treatment, and resolved slowly over a median of 46 (IQR 32.5–70) days.
Liver biopsies were obtained in only four of 25 (16 %) subjects adjudicated as probable or higher and, as is typical with idiosyncratic DILI, no pathonomonic features were evident [
17].
A number of patients were re-challenged with tolvaptan after elevated ALT values subsided. Approximately half of these subjects were able to tolerate the drug when it was re-introduced, suggesting that a form of adaptation or drug tolerance occurs. However, the other half experienced rapid ALT elevations upon re-exposure (positive re-challenge) and tolvaptan treatment had to be permanently withdrawn. The more rapid recurrence of injury upon re-exposure may indicate an adaptive immune mechanism and studies are underway to search for possible human leukocyte antigen (HLA) risk alleles as have been demonstrated for delayed DILI caused by other agents, including ximelagatran, lumiracoxib, and lapatinib [
18‐
20].
Following a recommendation from the TEMPO Steering Committee to increase the frequency of monitoring to monthly, no additional Hy’s Law cases have been identified to date (
n = 1275 subjects exposed for ≥18 months) (Table
1), further lowering the incidence of potential liver failure to approximately 1:4000. It should be noted that since liver chemistry monitoring was relatively infrequent in TEMPO 3:4 and its open-label extension, more frequent monitoring is expected to further lower the risk of liver failure. In some cases, liver injury did progress for weeks after stopping drug treatment, followed by slow resolution, and it seems unlikely that the risk of serious liver injury could be eliminated solely by more frequent monitoring. While no additional Hy’s cases have been identified since implementation of a comprehensive monitoring program consisting of monthly liver enzyme testing went into effect, the number of additional patients treated does not provide sufficient power to eliminate the possibility of another Hy’s case from occurring.
Of note, there was no evidence of a similar liver safety signal or signature presentation observed among subjects with cirrhosis, congestive heart failure, or hyponatremia exposed to tolvaptan in non-ADPKD clinical trials. This was a surprising finding, as the expectation might be that patients with liver disease or severe congestive heart failure would be more susceptible to DILI. The explanation is unlikely due to sample size, since the number of subjects treated for at least 14 months in the non-ADPKD populations (
n = 589) was not grossly different from the corresponding number of ADPKD subjects (
n ~ 860). Although the dose of tolvaptan received by the non-ADPKD population (single daily dose up to 60 mg/day) was lower than that received by most subjects in the ADPKD trials (split daily dose up to 120 mg/day), the overall exposure for most chronic heart failure subjects was similar to subjects receiving a split dose of 45/15 mg/day in the ADPKD population (data on file). It is possible that patients with ADPKD may be more susceptible to tolvaptan-associated liver injury than other patient populations due to an unknown feature of disease pathology. In this regard, it should be noted that the most common extra-renal manifestation of ADPKD is polycystic liver disease, which is characterized by the presence of multiple hepatic cysts originating from biliary ducts and peribiliary glands [
21]. In the CRISP (Consortium for Radiologic Imaging Studies of Polycystic Kidney Disease) study, hepatic cysts were observed in 58 % of 15- to 24-year-olds, 85 % of 25- to 34-year-olds, and 94 % of 35- to 46-year-olds with ADPKD [
22]. While imaging to document the presence of hepatic cysts was not included in the study protocols, more than 60 % of subjects in TEMPO 3:4 had liver cysts as determined by medical history or a retrospective review of their study MRIs. While it was not possible to correlate the presence of hepatic cysts with susceptibility to tolvaptan-associated liver injury in this study, 76 % of subjects adjudicated as probable or higher in TEMPO 3:4 and TEMPO 4:4 demonstrated some degree of hepatic cyst burden. It should also be noted that 18 of 30 (60 %) of subjects adjudicated as probable or higher and all three Hy’s law cases occurred in women, but it remains unknown whether this reflects true sex-specific differences.
Several risk-management strategies seem plausible for mitigating the potential for irreversible liver damage in the setting of long-term tolvaptan treatment for ADPKD, including more frequent liver testing at monthly intervals through the window of observed susceptibility. In addition, a personalized medicine approach might be feasible. Targeted genetic studies are currently underway to examine the potential role of the
HLA genotype, as are studies to determine if subjects with
PKD1 or
PKD2 have similar risks for hepatocellular injury [
23]. Alternatively, the observed liver injury may have a metabolic or mechanistic etiology. To address this possibility, drug-metabolizing enzymes, drug transporters, stress response proteins, mitochondrial markers, extracellular vesicle trafficking, and biliary disposition are all being examined. Finally, large-scale screening methodologies are being employed to detect other potential biomarkers that might be used to identify the rare ADPKD patient susceptible to tolvaptan-mediated liver injury.
6 Conclusions
In the pivotal TEMPO 3:4 clinical trial and its open-label extension, an imbalance in the number of subjects experiencing serum ALT elevations exceeding 3 × ULN was observed. The risk of liver failure in ADPKD patients receiving long-term tolvaptan therapy was estimated to be approximately 1:4000, with the latency of onset occurring primarily between 3 and 18 months of receiving therapy. Among the positively adjudicated cases, there were no reports of liver failure and all subjects experiencing hepatic injury recovered. Of note, the liver safety risk for tolvaptan-treated subjects in the ADPKD population was not evident in the tolvaptan-treated subjects in the non-ADPKD population, possibly due to enhanced susceptibility in the ADPKD population.
To further reduce the risk of liver injury in patients receiving long-term tolvaptan, we recommend frequent monitoring of liver function tests.
Compliance with Ethical Standards
All ADPKD and non-ADPKD clinical trials were supported by Otsuka Pharmaceutical Development & Commercialization Inc. All trials were conducted in compliance with the protocol the International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH) Good Clinical Practice (GCP) guideline and all other applicable regional regulatory requirements. Each trial site was approved by their local institutional review board or ethics committee according to regional requirements. Written informed consent was obtained from all subjects prior to initiation of any procedure being performed.
Each author contributed important intellectual content during manuscript drafting or revision and accepts accountability for the overall work by ensuring that questions pertaining to the accuracy or integrity of any portion of the work are appropriately investigated and resolved.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.