Is Bcl-2 a valid target in the treatment of indolent non-Hodgkin lymphoma?

Dysregulation of apoptosis through alterations in the B-cell lymphoma/leukemia 2 (Bcl-2) family of proteins is a hallmark of the pathophysiology of indolent non-Hodgkin lymphomas (NHLs). The most common indolent NHL, follicular lymphoma (FL), has a t(14;18) translocation in about 90% of cases, leading to profound overexpression of Bcl-2 due to juxtaposition of BCL-2 with the immunoglobulin heavy chain (IgH) promoter. The rarer, often more aggressive, mantle cell lymphoma (MCL) does not have a BCL-2 translocation, but nonetheless typically expresses high levels of Bcl-2. Why is the Bcl-2 family so fundamental to the pathogenesis of indolent NHL?

The answer may lie in the fact that the Bcl-2 family governs a fundamental cellular process: the intrinsic pathway of mitochondrial apoptosis. At the heart of the Bcl-2 family are the anti-apoptotic proteins such as Bcl-2 itself, as well as the related proteins Mcl-1 and Bcl-xL. These anti-apoptotic proteins bind the key pro-apoptotic proteins Bax and Bak, thereby inhibiting their oligimerization and putting the brakes on apoptosis. Activator and sensitizer pro-apoptotic proteins within the Bcl-2 family are known as BH3-only proteins, so named because they contain only the BH3 (Bcl-2 homology 3) domain found in other Bcl-2 family members. Activator BH3-only proteins include Bim and Bid, which directly activate Bax and Bak, and sensitizer BH3-only proteins such as Bad and Noxa that release Bax and Bak from the anti-apoptotic proteins and thereby indirectly promote apoptosis. The dynamic balance of pro- versus anti-apoptotic proteins is a key to determining the fate of a cell [1].

Given the Bcl-2 family's essential function in the pathobiology of indolent NHL, one would anticipate that antagonizing anti-apoptotic proteins would take the brakes off the intrinsic pathway of apoptosis and potently induce cell death in these malignant cells. Several agents have been touted to bind to Bcl-2 family members and mimic the normal biology of the BH3 domains of BH3-only proteins, and thus are known as BH3 mimetics. One such agent is obatoclax, which in vitro binds not only Bcl-2, but also Mcl-1 and Bcl-xL. It has been shown to cause dose-dependent cell death ex vivo in cells derived from patients with both untreated and relapsed/refractory FL [2]. In preclinical models of MCL, obatoclax can induce apoptosis as a single agent. Additionally, the combination of bortezomib and obatoclax in MCL cell lines and ex vivo primary MCL cells leads to synergistically increased cytotoxicity compared to either agent alone [3].

Despite these pre-clinical data, obatoclax's clinical activity has proven to be inconsistent. In this issue, two early phase clinical trials by Goy and colleagues explore the use of the BH3-mimetic obatoclax in newly diagnosed FL [4] and relapsed/refractory MCL [5]. The first trial is a small, sequential phase II study of obatoclax as a single agent followed by combination with rituximab in patients with previously untreated FL [4]. After 12 weeks, no objective responses were seen with single-agent obatoclax. When rituximab was combined with obatoclax, objective responses were seen in five patients (one complete response [CR] and four partial responses [PRs]), and the remaining two patients had stable disease at 24 weeks. The study was terminated prior to documentation of disease progression. All seven patients experienced neuropsychiatric adverse effects that were grade 1–2, associated with the obatoclax infusion, and resolved hours after completing the infusion. Three patients experienced grade 1–2 memory impairment that resolved after obatoclax was held for several days. While the number of patients treated was small, the data suggest that this combination does not have an advantage over rituximab monotherapy in FL.

The second trial is a phase I/II study of obatoclax and bortezomib in patients with relapsed/refractory MCL [5]. A total of 17 patients were treated at the dose level chosen for phase II expansion, until the study was closed prematurely due to slow enrollment and lack of a clear efficacy signal. Objective responses (two CRs, one CR unconfirmed [CRu] and two PRs) were seen in 22% (estimated median PFS 5.2 months), and an additional eight patients had stable disease for a median of 4 months. Overall, treatment was reasonably well tolerated, although 20% of patients experienced grade 3 or 4 myelosuppression. As in the FL study, most patients experienced grade 1–2 central nervous system (CNS) toxicity that was managed by dose-reduction or delay of obatoclax. Peripheral neuropathy, a well-described side effect of bortezomib, did not appear to be enhanced in combination with obatoclax. Overall, this combination of agents was tolerable at the dose levels and schedule chosen, but the combination did not appear to be more effective than historical data with bortezomib alone.

Why was obatoclax not as effective in these studies as had been hoped based on promising preclinical data? As the authors suggest, it is possible that obatoclax does not antagonize Bcl-2 as strongly in vivo as it does in vitro. Perhaps tumor cells in vivo are protected by microenvironmental factors that would require a higher dose of obatoclax than can be achieved due to the prominent neuropsychiatric side effects that were observed. Although it is challenging to find a pharmacodynamic biomarker for Bcl-2 antagonism, obatoclax purportedly also inhibits Bcl-xL, which should lead to substantial thrombocytopenia, since platelets depend on Bcl-xL for their survival [6]. The fact that thrombocytopenia was not a more prominent adverse event in either of these studies suggests that obatoclax is not adequately antagonizing Bcl-xL, and therefore is also unlikely to be hitting Bcl-2 effectively. At a more fundamental level, it is also possible that obatoclax has targets outside the Bcl-2 family and kills cells in vitro through alternative mechanisms. For example, preclinical work has demonstrated that obatoclax can kill cells in the absence of Bax and Bak, which would be required for apoptosis to occur through the intrinsic pathway [7]. This calls into question the extent to which the cytotoxicity of obatoclax has anything to do with its ability to bind Bcl-2. If the drug killed mainly through alternative mechanisms, one would not expect it to have a meaningful therapeutic window in Bcl-2 dependent diseases such as indolent NHL.

The shortcomings of obatoclax do not exclude the possibility that a molecule that effectively targets Bcl-2 in vivo will be helpful in treating indolent NHL. Other BH3 mimetics in development, such as those originally developed by Abbott Laboratories, appear to more convincingly target Bcl-2 family members in vivo. ABT-737 and navitoclax (ABT-263) are first-generation BH3 mimetics that bind Bcl-2 and Bcl-xL. The orally bioavailable navitoclax was taken forward in clinical studies, and in a phase I study in relapsed/refractory lymphoma showed promising efficacy: 21 of 46 patients with evaluable lymphadenopathy had a reduction in tumor size, including six of the 16 patients with FL [8]. Grade 3 or 4 thrombocytopenia was common (52.7% of patients) due to on-target effects of inhibiting Bcl-xL, suggesting that the drug was hitting its targets.

To get around the thrombocytopenia issue, a second-generation, Bcl-2 selective antagonist ABT-199 is being developed now. Early interim analyses of a phase I study of ABT-199 as a single agent in replaced/refractory lymphoid malignancies were recently presented. In another indolent lymphoma with high Bcl-2 expression, chronic lymphocytic leukemia (CLL), ABT-199 had impressive activity in a heavily pretreated population with high-risk disease, with an 84% overall response rate, including a 23% CR rate, and some patients achieving minimal residual disease negativity [9]. The efficacy signal in MCL has also been strong, with 9/11 patients (82%) responding, including one patient with a CR [10]. In FL, only 3/11 (27%) of patients have responded thus far, although several of these patients were treated at lower doses of ABT-199 in early dosing cohorts. Grade 3/4 thrombocytopenia has only occurred in 11% of patients, and has not been dose-dependent, making it unlikely to be related to the drug in the cases where it has been seen. Although ABT-199 does appear to be promising, the fact that a higher response rate has not been observed in FL raises important questions about the future of targeting Bcl-2 in indolent NHL. Can high enough drug levels be achieved to adequately antagonize the massive levels of Bcl-2 that are expressed? Even if Bcl-2 function is effectively decreased, does resistance arise through the microenvironment and/or up-regulation of other anti-apoptotics such as Mcl-1? Do malignant cells exposed to prolonged Bcl-2 antagonism develop somatic mutations in Bcl-2 that confer resistance?

The studies by Goy and colleagues provide helpful information about the clinical activity of obatoclax in indolent NHL, but they do not directly provide mechanistic insights to help us begin to answer these important questions. A key to improving our understanding of why drugs such as obatoclax have disappointing activity is to incorporate innovative pharmacodynamic correlative studies into future trials to help understand the mechanisms of how these agents kill tumor cells and how drug resistance arises. This information may help inform the design of trials that rationally combine BH3 mimetics with other targeted therapies to maximize the efficacy of these agents in indolent NHL. Incorporating novel correlative studies into these trials will provide the most efficient pathway forward to help translate our understanding of the biology of Bcl-2 dysregulation into more effective therapies for patients with indolent NHL.

Potential conflict of interest:

Disclosure forms provided by the authors are available with the full text of this article at www.informahealthcare.com/lal.