Hairy Cell Leukemia—New Genes, New Targets

Abstract

Hairy cell leukemia (HCL), a B cell malignancy comprising 2 % of all leukemias, has become quite exciting recently with regard to the development of new targets for therapy. This review will focus on advancements made within the past 1–2 years in targeted therapy for this disease. These advances may be grouped into two very difference categories, namely targeting of CD22 with the recombinant immunotoxin moxetumomab pasudotox, and targeting of the mutated BRAF component of the MAP kinase pathway. Moxetumomab pasudotox in phase I testing was recently reported to be associated with an overall response rate of 86 % and a complete remission (CR) rate of 46 % in 28 patients with relapsed and refractory HCL. Many of the CRs are without minimal residual disease (MRD). Severe or dose limiting toxicity was not observed on this trial, but a completely reversible and largely asymptomatic form of grade 2 hemolytic uremic syndrome occurred in two patients during retreatment. This agent has commenced phase III multicenter testing to validate its phase I results. An extensive number of studies have documented the V600E mutation in nearly all HCL patients, but not in similar hematologic malignancies. The thymidine kinase inhibitor vemurafenib, which inhibits the V600E mutant of BRAF, was reported to induce a CR in multiply relapsed and refractory HCL, with nearly complete clearing of MRD. One additional partial and one additional complete remission were subsequently reported.

Introduction

Hairy cell leukemia (HCL) is an indolent B cell malignancy comprising 2 % of leukemias, which presents with pancytopenia, splenomegaly, and malignant cells containing cytoplasmic projections [1]. Based on updated incidence data for leukemia [2, 3], there are approximately 1,000 new cases per year in the US and 1,600 new cases per year in Europe. The prevalence of HCL is not well studied, but is thought to greatly exceed the yearly incidence. This is based on the introduction of excellent life-prolonging therapies a quarter of a century ago, and the realization that most patients carry minimal residual disease (MRD) after therapy and many relapse from complete remission (CR) after years of observation. In recent years, HCL has become one of the most interesting leukemias with respect to new targets and therapies, which will be the subject of this review.

Diagnosis of HCL

The clinical presentation of HCL includes fatigue, right upper quadrant pain, bleeding and infection, due to cytopenias and splenomegaly [4]. The median age of onset is 55 years and the male to female ratio is 4.5 [5]. Typical blood counts include leucopenia, anemia, thrombocytopenia, neutropenia and monocytopenia [6]. Hypocholesterolemia is another laboratory abnormality often present when HCL patients are diagnosed [7]. The laboratory diagnosis of HCL was originally made by cell morphology, with cells demonstrating cytoplasmic projections, and bone marrow biopsy showing the typical ‘fried egg’ appearance due to abundant cytoplasm [8]. Spleen morphology shows diffuse involvement of red pulp [9]. Flow cytometry of the blood or marrow aspirate combined with immunohistochemistry of the bone marrow biopsy is now most often used for diagnosis. HCL expresses B cell antigens CD19, CD20 and CD22, and also CD11c, CD103 and CD25 [10–14]. The most brightly expressed of these include CD20, CD22, CD11c, and sometimes CD25. Immunohistochemistry of the bone marrow biopsy shows TRAP, DBA.44, and Annexin A1 (Anxa1) positivity in classic HCL [15–17].

Diagnosis of HCL variant (HCLv)

A poor-prognosis, treatment-resistant variant of HCL, called HCLv, was first described in 1980 and comprises about 10 % of the cases of HCL [9, 18]. The median age of diagnosis is 71 years and the male–female ratio is less than 2 [19]. Rather than HCL-related cytopenias, patients with HCLv have lymphocytosis and unlike patients with classic HCL, HCLv may present with adenopathy, and more severe splenomegaly [9, 19, 20]. Splenic infiltration into the red pulp is similar to that in HCL [9]. Flow cytometry shows bright positivity for CD20, CD22 and CD11c, but differs in several key antigens, including absence for CD25 and usually CD123 [12–14, 19]. While CD103 may also be absent in a minority of patients with HCLv [9], the presence of CD103 and absence of CD25 is a key combination that argues for HCLv and against the diagnosis of splenic marginal zone lymphoma (SMZL) [14, 21]. Annexin A1 was reported absent in ten of ten cases [17]. A molecularly defined variant of HCL expressing the IGHV4–34 rearrangement was described in a population enriched for being relapsed and refractory, comprising 10 % and 40 % of the patients with immunophenotypically defined HCL and HCLv, respectively [22]. These patients present clinically with more lymphocytosis and fewer cytopenias, like the HCLv patients.

Treatment of HCL and HCLv with Interferon and Purine Analogs

Interferon-alpha was the first highly effective systemic therapy discovered for HCL, first reported with 30 % complete remission (CR) and 56 % partial response (PR) rates [23], In multiple follow-up studies, CR rates were somewhat lower, with PR rates of 69–87 %, and relapse would occur soon after stopping the drug [24]. Of 52 patients with HCLv, only two (14 %) responded to interferon-alpha [19]. The purine analogs pentostatin and cladribine proved to be far superior to interferon-alpha in the treatment of HCL. Pentostatin, which was tested against interferon-alpha in a randomized trial [25, 26], achieves CR rates of 72–89 % in HCL and 8-year disease-free survivals of 76 % [25–31]. Cladribine is usually administered in a single 5–7 day course, while pentostatin is given biweekly, usually for 3–6 months. Cladribine achieves similar CR rates, 75–87 %, and 7-year disease-free survivals of 60–66 % [32–40]. However, in HCLv, of 35 patients from six retrospective studies of 3–15 patients each, cladribine achieved only 9 % CRs and 40 % PRs [19, 22, 41–44], and pentostatin in 15 with HCLv achieved 0 % CR and 54 % PR [19]. Due to major differences between HCL and HCLv in diagnosis and outcome, HCLv was classified in 2008 by the World Health Organization (WHO) as an entity within ‘Splenic B cell lymphoma/leukemia, unclassifiable’, separate from classic HCL [45, 46]. In the molecularly defined IGHV4–34 variant of HCL/HCLv, initial cladribine was poorly active, with 0 % CR and 15 % PRs in 13 patients [22].

Need for Investigational Approaches

Despite the outstanding results with purine analogs in classic HCL, realities regarding long-term outcome confirm the need for investigational targeted approaches. Else et al. reported a median relapse-free survival (RFS) of 16 years, including all 233 HCL patients receiving first-line purine analog therapy, and RFS dropped to 11 and 6.5 years with subsequent lines of purine analog [47]. From the 358-patient Scripps Clinic database, 19 patients were followed in continuous CR after a median of 16 years from cladribine, and 47 % were still MRD-free [48]. Thus, while CR rates and durations are excellent with initial and repeat courses of purine analog, and a minority of patients may remain MRD-negative and hence ‘cured’, the majority may continue to harbor HCL and require retreatment until purine analog refractory. Since this process may take a median of over 35 years and we are still only 25 years since the introduction of purine analogs, patients diagnosed young enough and living long enough will continue being added to the growing population of patients who are refractory to treatment. Moreover, initial and certainly repeated courses of purine analogs are toxic, with neuropathy and long-term damage to CD4+ T-cells documented even to a single course [49–51].

Monoclonal Antibody Approaches for HCL

The most common non-chemotherapy targeted approach used today for HCL is rituximab, the anti-CD20 murine-human chimeric monoclonal antibody (Mab) that kills cells by inducing apoptosis, and antibody-dependent and complement-dependent cytotoxicity [52]. HCL cells are strongly CD20 positive [11]. Beginning in 1999, case reports documented efficacy of rituximab in HCL [53–57], and CR rates among six reported studies (10–25 patients each, total of 97, getting usually 4–8 weekly doses) varied from 10 to 54 % [58–63]. However, in 48 patients from five studies who needed treatment because of cytopenias (hemoglobin <10 g/dL, platelets <100,000/mm³, or neutrophils <1,000/mm³) and who had at least one prior purine analog, there were nine (19 %) CRs and ten (21 %) PRs [58–62]. In the largest single trial enrolling 24 patients, all of whom needed treatment for cytopenias and had prior purine analog, there were three (13 %) CRs and three (13 %) PRs [60]. Thus, rituximab efficacy in HCL appears limited in relapsed patients who need treatment because of cytopenias, possibly due to limited antibody penetration into large clumps of HCL cells, or acquired resistance. Other MAbs, including alemtuzumab, have been reported for HCL, but efficacy has been limited in these anecdotal reports [64, 65]. In the past year, progress has been made targeting HCL using recombinant immunotoxins, particularly those targeting CD22, and inhibiting mutated BRAF. This review will now focus on these two approaches.

Introduction to Immunotoxins

To target drug resistant cells using surface molecules as antigen targets, but without relying on mechanisms of Mab-induced killing, immunotoxins were created by linking MAbs to protein toxins. Immunotoxins kill cells by binding to a cell, internalizing, and trafficking a toxin fragment to the cytoplasm, where it catalytically inhibits protein synthesis, killing the cell by apoptosis. Both plant and bacterial toxins have been shown to kill a cell with a single molecule in the cytoplasm [66–68]. The bacterial toxins diphtheria toxin (DT) and Pseudomonas exotoxin (PE) are particularly useful since they are made with a binding and catalytic domain in single-chain form [69]. A recombinant toxin can be made by substituting a cancer cell-binding ligand for the native binding domain of the toxin. The 38 kDa truncated form of PE missing its binding domain is called PE38. Recombinant immunotoxins contain the variable domains of the Mab (VH and VL) connected together by either a (G₄S)₃ peptide linker (single-chain) or an engineered disulfide bond (disulfide-stabilized), and one of the variable domains is fused to the toxin [70, 71]. Although the disulfide-stabilized recombinant immunotoxins are composed of two chains disulfide-bonded together, the immunotoxin is nevertheless recombinant, since the disulfide bond forms during renaturation without chemical conjugation. Recombinant immunotoxins targeting CD25 include LMB-2 and those targeting CD22 include BL22 (CAT-3888) and its affinity-matured version, moxetumomab pasudotox (HA22 or CAT-8015).

Activity of LMB-2 on HCL

LMB-2 contains the VH and VL domains of the anti-CD25 Mab anti-Tac as a single-chain Fv, and VL is fused to PE38 [72, 73]. LMB-2 in phase I testing achieved 100 % responses in four patients with HCL (1 CR, 3 PR) [74, 75]. However, since CD25 to which LMB-2 binds is absent on the 10 % of HCL patients with HCLv cells, and some HCL patients have HCL cells which undergo loss of CD25, recombinant immunotoxin targeting of HCL shifted to CD22, a more ubiquitous antigen for HCL [76]. LMB-2 is continuing testing in a phase II trial for HCL patients ineligible for moxetumomab pasudotox.

CD22 as a Target in HCL

CD22 is a 140-kD phosphoglycoprotein member of the Ig-superfamily, also called sialic acid binding Ig-like lectin 2 (Siglec-2), that functions as a B cell receptor signal transduction modifier and a cellular adhesion molecule [77]. CD22 is expressed on mature B cells [78] and much more intensely on 100 % of hairy cells [11, 14, 76, 79, 80].

Development of BL22 for HCL

BL22, also called CAT-3888, is an anti-CD22 recombinant immunotoxin containing the variable domains of the anti-CD22 monoclonal antibody (Mab) RFB4, fused to PE38. Preclinical testing validated its ability to target CD22+ cells [81, 82]. In phase I testing of 31 HCL patients, there were 19 (61 %) CRs and six (19 %) PRs [83, 84]. A completely reversible form of hemolytic uremic syndrome (HUS) was observed in four (13 %) patients, but only during retreatment with a second or third cycle. However, most (58 %) CRs were achieved with one cycle, suggesting that retreatment was not always necessary or advisable. Thus in phase II testing, patients were retreated if cytopenias persisted after cycle 1, and received a lower dose level. One cycle of BL22 achieved CR in 25 % of 36 patients. After retreatment of 56 % of patients, the CR rate was 47 % with overall response rate (ORR) 72 % [85]. Compared to the phase I trial, the dose-limiting HUS rate was less than half (6 % vs. 13 %) and the immunogenicity rate (production of neutralizing antibodies against the toxin) less than one-third (11 % in phase II vs. 35 % in phase I). It was concluded that BL22 is active in HCL, despite prior purine analog treatment and resistance, and that patients needed to be followed closely for HUS, to make sure it would completely reverse if it were to occur.

Development of Moxetumomab Pasudotox

To allow a higher percentage of bound immunotoxin to be internalized by CD22 rather than disassociating from this antigen, the off-rate of BL22 was lowered by mutagenesis at the third complementarity determining region of the heavy chain. The resulting anti-CD22 recombinant immunotoxin moxetumomab pasudotox (previously called CAT-8015 or HA22) contained three amino acid mutations and had 14-fold higher affinity and lower off-rate to CD22, leading to improved cytotoxicity [86–88].

Dose-Escalation Phase I Testing of Moxetumomab Pasudotox in HCL

To determine a safe and effective dose of moxetumomab pasudotox, 28 HCL patients with relapsed/refractory HCL disease were treated [89••]. Patients received 1–16 (median 4) cycles each, with cycles spaced 4 weeks apart, for a total of 114 cycles, with each cycle comprised of three doses every other day (QOD x3). Three patients each received 5, 10, 20, and 30 ug/Kg QOD x3, four received 40 ug/Kg QOD x3, and the remaining 12 patients received 50 ug/Kg QOD x3. No dose limiting toxicity (DLT) was observed at any dose level, but two patients had reversible laboratory abnormalities consistent with grade II HUS, one after three cycles of 30 and one after five cycles of 50 ug/Kg QOD x3 [89••]. Common toxicities observed, including hypoalbuminemia, edema, hypotension and proteinuria, were often related to low-grade capillary leak syndrome (CLS). Responses were achieved at all doses levels, with CRs as low as 10 ug/Kg QOD x3. Of the 28 evaluable patients, the ORR was 86 % with 13 (46 %) CRs. Due to the high CD22 antigen density, higher tumor burden was associated with lower peak plasma levels, and due to decreasing tumor burden between days 1 and 5, plasma levels rose significantly [89••]. ELISA was used to quantitate immunogenicity leading to binding antibodies, and was positive in 17 (65 %) of 26 patients evaluated after a median of two cycles. Cytotoxicity assays used to quantify neutralizing antibodies were positive in 10 (38 %) of 26 evaluable [89••], usually by cycle 3. While immunogenicity did not prevent CR, which required two to five cycles of moxetumomab pasudotox, consolidation cycles could not be given in patients after immunogenicity was detected. Otherwise, up to 2 extra cycles cloud be given after achievement of CR. Five patients achieving CR received ten consolidation cycles, and the other eight CRs had either immunogenicity (n=6) or grade II HUS (n=2). Of 13 achieving CRs, 83 % remained in CR at a median of 29 months. While response was not related to prior purine analog, CR rate was adversely affected by prior splenectomy, with zero of seven asplenic patients achieving CR compared with 13 of 21 with intact spleens (p=0.007). Patients without spleens still responded, as six (86 %) of these seven had PR. Failure to achieve CRs with moxetumomab pasudotox after prior splenectomy was probably related to more advanced disease, and possible increased bone marrow infiltration with HCL in the absence of a spleen catching malignant cells. Thus, moxetumomab pasudotox showed clinical response in relapsed/refractory HCL, and its safety profile supported further clinical development.

Development Plan for Moxetumomab Pasudotox in HCL

To confirm the activity and safety of moxetumomab pasudotox for relapsed and refractory HCL, a multicenter phase III pivotal trial has opened enrolling patients with at least two prior courses of purine analog, requiring treatment due to either cytopenias or painful splenomegaly. Patients who were refractory to their first course of purine analog, defined as response lasting under a year, may be enrolled after just one prior purine analog and at least one prior course of rituximab. This trial will not require testing of neutralizing antibodies for eligibility, since patients in phase I testing did not have adverse events as a result of immunogenicity, and since less than one-fourth of patients would be expected to have pre-existing antibodies to the agent. Since cumulative toxicity was not observed in phase I testing, this trial is designed to not limit repeat cycles in responding patients, and to allow patients achieving CR without detectable MRD two consolidation cycles. It is notable that of the 12 patients reported to receive the maximum dose, six (50 %) achieved CR, and three of these patients had CR without any detectable MRD. The responses of these three patients are ongoing for 41–54 (median 48) months without detectable MRD, despite repeat bone marrow and blood studies. Further studies and follow-up will be needed to determine if targeting an HCL surface antigen in this manner could eradicate the malignant clone.

Introduction to BRAF in HCL

To search for a treatable target in HCL that would be important for leukemogenesis, Tiacci et al. performed whole exome sequencing comparing a patient’s purified HCL and non-malignant cells [90••]. This study yielded five candidate mutated genes, including BRAF and four others (CSMD3, SLC5A1, CNTN6, and OR8J1). While the latter four lacked biologic significance, BRAF was suspected as an important gene, since it is an important member of the mitogen activated protein kinase (MAPK) pathway that influences cell proliferation [91]. BRAF was noted to be the most frequently mutated protein kinase in cancer [92], and the BRAF mutation found, c.1799 T>A, encoding the amino acid substitution V600E, is the most common BRAF mutation found in melanoma and other cancers [92, 93]. In this pathway, BRAF, which is activated by RAS, activates MEK by phosphorylation, which in turn activates ERK by phosphorylation, and the final result is increased cell proliferation [91, 92]. Tiacci et al. also noted that a previous study by Kamiguti et al. from 2003 implicated the MAPK pathway in the pathogenesis of HCL, and demonstrated that inhibition of this pathway led to HCL death [94]. In the study by Tiacci et al., 47 out of 47 other patients were studied and found to have HCL cells containing V600E, compared to 0 of 195 patients with other malignancies, confirming the importance of this mutation in the pathogenesis of HCL.

Extensive Studies of BRAF Mutations in HCL and Other Hematologic Malignancies

Many studies have been carried out to define the incidence of BRAF V600E expression in HCL and other hematologic malignancies, all listed in Table 1. In their original report, Tiacci et al. performed PCR and direct DNA Sanger sequencing, and found 100 % of HCL vs. 0 % of other B cell malignancies expressed V600E [90••]. Boyd et al. used a high resolution melting analysis (HRMA) capable of detecting 5–10 % HCL cells in a sample, and reported V600E in 100 % of 48 HCL samples compared to 0 % of 114 other malignancies, including one multiple myeloma which expressed BRAF D594N [95]. Blombery et al. also used HRMA of similar sensitivity and found similar results, including a splenic marginal zone lymphoma (SMZL) patient expressing K600E [96]. We preformed BRAF V600E testing using a targeted pyrosequencing [97] employing COLD-PCR conditions [98], which allow for enrichment of mutant V600E sequences compared to native BRAF DNA. This technique was capable of detecting V600E+HCL comprising 2–3 % of the sample, although samples with <10 % HCL cells were excluded. Unlike other studies, most of the 53 HCL patients used in our study were multiply relapsed. The IGHV4–34 rearrangement, a marker of poor prognosis whether flow cytometry is consistent with HCL or HCLv [22], characterized five of 53 HCL patients and eight of 16 HCLv patients [99•]. Not only were all HCLv patients negative for V600E, including the eight IGHV4–34+ HCLv patients, but all five IGHV4–34 HCL patients were V600E negative (Table 1). One of six patients with unknown IGHV type was BRAF V600E negative, so this patient could have been IGHV4–34+ as well. It was also found that five of 42 classic HCL patients of known IGHV type were negative for BRAF V600E, suggesting that these patients require further study and characterization. Three other studies employed pyrosequencing, documenting 100 % V600E positivity in HCL [100–102]. Allele-specific PCR was used to detect V600E+HCL in five studies [103–106], including a follow-up study by Tiacci et al. [104] that reported detection of as few as 0.1 % HCL cells using imaging on agarose gels. Other studies have employed RQ-PCR [107, 108], DNA Sanger sequencing [109], bidirectional PCR sequencing [110], and immunohistochemistry using the VE1 MAb specific for BRAF V600E [111]. These studies confirmed 100 % V600E positivity in HCL with only three exceptions. Schnittger et al. reported two out of 117 patients negative for BRAF V600E, despite classic HCL immunophenotyping and absence of IGHV4–34 [108]. Langabeer et al. reported one case of classic HCL without BRAF V600E expression [105, 106]. Thus, with very rare exceptions, BRAF V600E is an extremely sensitive and specific marker in HCL, and represents an opportunity for therapy.

Table 1.

Studies of BRAF in HCL and other hematologic malignancies

First Author	Patients	Test	V600E	Ref
Tiacci et al.	48 HCL	DNA Sanger sequencing	48/48	[90••]
	22 SMZL		0/22
	16 SLLU		0/16
	21 CLL		0/21
	35 FL		0/35
	71 DLBCL		0/71
	18 MCL		0/18
	12 BL		0/12
Boyd et al.	48 HCL	HR Melting analysis	48/483	[95]
	22 CD103+ LDs		0/22
	43 SMZL		0/43
	39 MM		0/394
Blombery et al.	36 active HCL	HR Melting analysis	36/36	[96]
	5 nodal MZL		0/5
	10 SMZL		0/102
Xi et al.	42 HCL	Pyrosequencing	37/42	[99•]
	6 HCL, IGHV?		5/6
	5 HCL IGHV4–34		0/5
	16 HCLv		0/16
Verma et al.	12 HCL	Pyrosequencing	12/12	[100]
	4 HCLv		0/4
Laurini et al.	9 HCL	Pyrosequencing	9/9	[101]
	6 HCLv		0/6
	10 nodal MZL		0/10
	10 extranodal MZL		0/10
	10 PTLD		0/10
	11 PTCL		0/11
	12 ALCL		0/12
	10 LGL		0/10
Lennerz et al.	17 HCL	Pyrosequencing	17/17	[102]
	1 CD5+ HCLv		0/1
Arcaini et al.	62 HCL	Allele-specific PCR	62/62	[103]
	1 HCLv		0/1
	91 SMZL		0/91
	29 WM		0/29
	57 B-CLDs		2/571
Trifa et al.	309 MPD	Allele-specific PCR	0/309
	76 CML		0/76
	19 PMF		0/19
	57 AML		0/57
	36 MDS		0/36
Tiacci et al.	123 HCL	Allele-specific PCR	123/123	[104]
	61 SMZL		0/61
	11 HCLv		0/11
	7 SLLU		0/7
	31 CLL		0/31
	5 CD5neg BCU		0/5
Langabeer et al.	24 HCL	Allele-specific PCR	23/24	[105, 106]
	3 HCLv		0/3
Ewalt et al.	12 HCL	RQ-PCR	9/95	[107]
	20 CLL/SLL		0/20
	19 MCL		0/19
	22 nodal MZL		0/22
	2 SMZL		0/2
	7 LPL		0/7
Schnittger et al.	117 HCL	RQ-PCR	115/117	[108]
	16 HCLv		0/16
	9 AML		0/9
	7 MDS		0/7
	13 MPD		0/13
	14 CML		0/14
	4 CMML		0/4
	2 CEL		0/2
	1 T-ALL		0/1
	2 B-ALL		0/2
	3 FL		0/3
	8 MCL		0/8
	1 SMZL		0/1
	11 CLL		0/11
	20 NHL		0/20
Jebaraj et al.	138 CLL	DNA Sanger sequencing	4/138	[109]
	32 B-PLL		0/32
Ping et al.	288 AML	Bidirectional PCR sequencing	0/288	[110]
	84 MDS		0/84
	84 MPD		0/84
	122 CML		0/122
	112 B-ALL		0/112
	17 T-ALL		0/17
	47 CLL		0/47
	26 MM		0/26
Andrulis et al.	32 HCL	VE1 Immunohistochemistry	32/32	[111]
	2 HCLv		0/2
	8 SMZL		0/8
	6 SLLU		0/6
	4 B-NHL		0/4
	26 Plasmacytoma		0/26
	34 FL		0/34
	41 DLBCL		0/41
	25 PMBL		0/25
	17 MCL		0/17
	16 extranodal MZL		0/16
	46 HL		0/46
	21 CLL		1/21

ALL acute lymphoblastic leukemia; AML acute myelogenous leukemia; ALCL, anaplastic large cell lymphoma; B-CLD B cell chronic lymphoproliferative disorders; BL Burkitt’s lymphoma; CD5neg BCU CD5-negative B cell neoplasms, unclassifiable; CLL/SLL chronic lymphocytic leukemia/small lymphocytic lymphoma; CEL chronic eosinophilic leukemia; CML chronic myelogenous leukemia; CMML chronic myelomonocytic leukemia; DLBCL diffuse large B cell lymphoma; FL follicular lymphoma, HCL hairy cell leukemia; HCLv HCL variant; HR High resolution; LDs lymphoproliferative disorders; LGL large granular lymphocyte proliferation; LPL lymphoplasmacytic lymphoma; MDS myelodysplastic syndrome; MM multiple myeloma; MPD myeloproliferative disorder; MZL marginal zone lymphoma; PLL prolymphocytic leukemia; PMF primary myelofibrosis; PTCL peripheral T cell cell lymphoma; PTLD post-transplantation lymphoproliferative disorder; RQ-PCR real time quantitative polymerase chain reaction; SMZL splenic marginal zone lymphoma; SLLU splenic lymphoma/leukemia, unclassifiable; WM Waldenstrom’s macroglobulinemia

¹Sanger sequencing negative, indicating V600E was present in a subclone

²K600E found in one with SMZL

³Silent GAT to GAC mutation in one patient at codon 594

⁴GAT to GTA at codon 594 (D594N) and silent GTG to GTA at codon 600 in one MM patient

⁵In addition, three HCL patients with <5 % tumor burden were negative by RT-PCR

Development of Vemurafenib for Melanoma and Other Cancers

The discovery of the V600E mutation in HCL was particularly exciting because of the rapid development of this target for therapy in melanoma and other malignancies. Mutant BRAF has been documented in ~40–60 % of malignant melanoma, 5–10 % of colorectal cancer, ~40 % of papillary thyroid carcinoma, ~2 % of adenocarcinomas of the lung, and a proportion of ovarian carcinomas, Langerhans histiocytosis, anaplastic thyroid cancers, biliary tract cancers, diffuse large B cell lymphomas, gastrointestinal stromal tumors, germ cell tumors, gliomas, adenocarcinomas of the small intestine and multiple myelomas [92, 112–122]. In 2011, Chapman et al. reported that vemurafenib, a thymidine kinase inhibitor which specifically inhibits BRAF containing the V600E mutation, showed improved overall survival (OS) compared to dacarbazine (DTIC) for V600E+melanoma, and also increased progression-free survival (PFS) [123]. The response rates of vemurafenib and DTIC were 48 % vs. 5 %, respectively. Grade 2–4 (moderate to life-threatening) toxicity of vemurafenib included arthralgia (21 %), photosensitivity rash (18 %), fatigue (13 %), and squamous cell carcinomas of skin (12 %). This data led to FDA approval of vemurafenib for V600E+advanced melanoma. One exciting feature of vemurafenib to the melanoma community is its activity in the central nervous system, for metastatic disease to both brain [124] and leptomeninges [125].

Activity of Vemurafenib in HCL

At this time, clinical trial data is unpublished, but several responses of HCL to vemurafenib have been recently reported. Dietrich et al. reported a 51-year-old male with refractory HCL and severe cytopenias to completely respond to vemurafenib 240–960 mg twice daily by 43 days of treatment, and continued for a total of 56 days [126••]. A small amount of HCL MRD remained in the bone marrow aspirate by flow cytometry. This patient showed no evidence of relapse 6 months after treatment [127]. A second patient was reported who received only 240 mg twice daily for 58 days, achieving a PR [128]. Treatment was complicated by several benign seborrhoeic keratosis lesions requiring surgical removal. A third patient was reported to achieve a CR by day 90 of vemurafenib 240 mg twice daily given from day 1 to 56 [129]. Thus, in anecdotal reports, vemurafenib is capable of achieving rapid response in BRAF V600E+ HCL, and short follow-up indicates that response persists after discontinuation of therapy. However, eradication of MRD by sensitive tests, such as flow cytometry of the bone marrow aspirate, has not been reported, and it is still unknown whether single agent BRAF inhibitor can lead to multi-year CR.

Development of Other Inhibitors of the MAPK Pathway

A second inhibitor of BRAF V600E, termed dabrafenib, was tested in a phase III randomized trial versus DTIC in patients with advanced melanoma, and showed a significant improvement in PFS from 2.7 to 5.1 months with hazard ratio 0.3, p<0.0001 [130]. Grade 2–4 toxicity included hyperkeratosis (~12 %), palmar-plantar hyperkeratosis (16 %), and squamous cell carcinomas and keratoacanthomas of skin (8 %), Because relapse despite BRAF inhibition of melanoma was rapid, inhibitors were developed to other components of the MAPK pathway. In 2012, Flaherty et al. reported that the MEK inhibitor trametinib compared to chemotherapy in metastatic melanoma resulted in increased PFS (4.8 vs. 1.5 months, p<0.001) and 6-month OS rate (81 vs. 67 %, p=0.01) [131]. Grade 2–4 toxicity included rash (27 %), diarrhea (6 %), fatigue (9 %), acneiform dermatitis (10 %), and hypertension (15 %). A second MEK inhibitor termed selumetinib (AZD6244, ARRY142886) was recently tested in a phase II trial in V600E+ melanoma [132]. This trial was restricted to patients with non-activated PI3K/AKT pathway, since none of ten patients with high phosphorylated-AKT (pAKT) expression responded. Three of five patients with low pAKT expression achieved tumor regression, one of whom achieved PR. Toxicity included rash, fatigue, and elevated liver function tests, some grade 3–4, and one patient who was responding but had grade 2 cardiac toxicity was taken off treatment before achieving PR.

Combined BRAF and MEK Inhibition

An important strategy in the development of MAPK inhibition is to combine BRAF and MEK inhibition simultaneously. An open label study was performed in metastatic melanoma, in which 85 patients were treated with different doses of dabrafenib plus trametinib, and 162 patients were then randomized to dabrafenib (150 mg) plus trametinib (1 or 2 mg) or to dabrafenib alone [133]. A trend was observed for fewer squamous cell carcinomas in the combined group with higher trametinib dose (150/2) compared to dabrafenib alone (7 % vs. 19 %, p=0.09), while the combination resulted in more fever (71 vs. 26 %). Dabrafenib-type toxicities were less frequent in 150/2 combination treatment compared to dabrafenib alone, such as hyperkeratosis (9 % vs. 30 %) and papilloma (4 % vs. 15 %), while trametinib-type toxicities were more frequent in the combined group, such as edema (29 % vs. 17 %), hypertension (9 % vs. 4 %), decreased ejection fraction (9 % vs. 0 %) and chorioretinopathy (2 % vs. 0 %). Median PFS was significantly longer with combination (150/2) than with dabrafenib alone (9.4 mo vs. 5.8 mo, p<0.001). One-year OS was 79 % for combination (150/2) vs. 70 % for dabrafenib alone, despite the fact that 80 % of the dabrafenib monotherapy arm crossed over to combination (150/2). Compared to the phase III trial of trametinib alone discussed above [131], it was felt that acneiform dermatitis, which is the most common dose limiting toxicity of trametinib, was reduced by combined (150/2) treatment (0 % vs. 8 %).

Conclusion

New Targets in HCL

For patients with multiply relapsed HCL after purine analog therapy, new therapies in development, including moxetumomab pasudotox targeting CD22 on the cell surface, and BRAF and/or MEK inhibitors, targeting the MAPK pathway, hold promise for efficacy without chemotherapy toxicity. However, the discovery of new targets for therapy in HCL is far from complete. In classic HCL, mechanisms of resistance to BRAF and/or MEK inhibition will require the discovery of methods to block escape pathways, and new targets that could be exploited. In HCLv, and in particular IGHV4–34+ disease, etiologic and potentially therapeutic targets remain to be discovered. Also, new agents like ibrutinib, which have demonstrated efficacy in several hematologic malignancies by BTK inhibition [134], are beginning clinical testing in both HCL and HCLv.