nbUVB radiation works by creation of DNA photoproducts, namely pyrimidine dimers, which leads to replication arrest or apoptosis. Its efficacy in MF is likely due to its ability to induce apoptosis in T cells (154). Many studies evaluating the effects of UVB photobiology have been carried out following irradiation of immune cells or keratinocytes in vitro or following UVR of murine models (155). In these settings, both release of pro-inflammatory and immunosuppressive cytokines have been observed following irradiation. However, the majority of studies have observed immunosuppressive effects of nbUVB, including altered antigen presentation, Langerhans cell depletion, decreased activity of NK cells (156, 157), and generation of Tregs (147, 158–161). Despite these negative immune effects, nbUVB is one of the most commonly used and effective skin-directed therapies for MF ( Table 3 ) (34).

PUVA is photochemotherapy consisting of taking the oral drug psoralen and then exposing the skin to UVA light. Psoralen forms DNA crosslinks following UVA exposure leading to DNA damage and apoptosis (162). Unlike nbUVB, the effect of PUVA has been studied in detail in an MF cohort leading to several major observations (163). The first is that PUVA can indeed reduce or eliminate malignant T cell as well as decrease benign T cells from the skin. However, this largely occurred in those with a “low” pre-treatment tumor burden (<10% tumor clone frequency, TCF, the percent of T cells that are malignant). In those with a “high” initial tumor burden (>20% TCF), clinical resolution of inflammation was independent of changes in total malignant or benign T cell numbers but due to a change in benign T cell populations based on characterization of the T cell repertoire by high throughout sequencing (HTS) of the T cell receptor (TCR). Bulk RNA sequencing coupled with this HTS TCR data suggested that PUVA may induce a turnover of benign T cell from T_h2 to T_h1 populations. Interestingly, a subset of benign T cells recruited post-PUVA treatment showed markers of TCR-dependent, antigen-specific activation-markers of anti-tumor T cells. Therefore, PUVA may lead to clinical responses by direct clearance of malignant T cells or indirectly via generation of an anti-tumor immune response. And, surprisingly, PUVA can lead to clinical responses through clearance of inflammation-producing benign T cells without reducing malignant T cells. Despite the later observation, PUVA has been successfully used in MF with long-term remissions possible ( Table 3 ) (35, 164, 165). A number of retrospective cohort studies have evaluated PUVA in combination with IFN-α, but given the lack of a comparator arm it is difficult to assess the full benefit of the addition (166–171). In one randomized clinical trial comparing IFN-α and PUVA versus PUVA monotherapy in CTCL stages I and II, there was not a statistically significant difference in complete remission, but the combination therapy had increased progression-free time (172).

ECP involves ex vivo treatment of a patient’s blood followed by retransfusion into the patient (173). ECP, typically used for MF/SS patients with a low blood burden of disease ( Table 3 ) (25, 36), exposes peripheral blood (buffy coat) to UVA radiation following photosensitization with 8-methoxypsoralen (8-MOP) (173). It has direct pro-apoptotic as well as immune modulating effects (174). During the ECP process, platelets adherent to the ECP plate, inadvertently coated with fibrin, engage monocytes and induce them to mature into dendritic cells capable of presenting tumor antigens to generate an anti-tumor immune response (175). While the type of apoptosis induced in leukemic malignant T cells has not been further evaluated, in 8-MOP treated melanoma cell lines, photoactivation induced an immunogenic cell death (176). While ECP has been shown to eliminate alloreactive T cells in GVHD, it has also been reported to induce the generation of Tregs, a potentially unfavorable immune response in CTCL (177). Understanding the effect of ECP on Treg populations in CTCL in the past has been challenging given the difficulty of distinguishing malignant T cells expressing Treg markers from benign Tregs (178). Additionally, in murine models, ECP impairs effector T cell functions (179). Several retrospective and some prospective cohort studies have suggested that combination of interferon or bexarotene with ECP may lead to improved responses, although further studies are needed (60).

PDT involves the combination of a photosensitizing compound, such as porphyrins (e.g. 5-aminolevulinic acid, ALA) or non-porphyrins (e.g. hypercin), with administration of light of a certain wavelength (180, 181). These photosensitizers may preferentially accumulate in the mitochondria, lysosomes, endoplasmic reticulum (ER), Golgi apparatus, and/or plasma membrane. The light then activates the sensitizer, generating reactive oxygen species (ROS) (182). ROS can be generated by physiologic signaling pathways in T cells leading to activation, differentiation, proliferation, or apoptosis (183). However, high levels of ROS generated by photosensitizers, especially those leading to ER stress, potently induce immunogenic cell death (181). Following PDT, pro-inflammatory cytokines IL-1 and IL-6 are also increased (184). While inflammatory apoptosis of malignant T cells is favorable, Tregs are not as suspectable to ROS (185). In addition, ROS can cause hyporesponsiveness of effector T cells and NK cells (185). Therefore, generation of ROS can be a double-edged sword.

PDT has shown promise for the treatment of MF. Case series evaluating the use of PDT date to the 1990’s (186). Initial studies used ALA, but this was later replaced by methyl-aminolevulinate (MAL), which is thought to have better specificity given its lipophilic properties (186). Despite encouraging responses ( Table 3 ) (26), the use of in office PDT for MF is impractical given the size of lamp and need for frequent treatments. Not surprisingly PDT in the past was mainly investigated in unilesional MF (186). Recently, a non-porphyrin photosensitizer, synthetic hypercin (SGX301), which is activated by visible fluorescent light in the 500-650mn range (yellow-red spectrum) has completed its pivotal phase 3 clinical study. In in vitro studies activated hypercidin caused greater apoptosis of malignant T cells over benign T cells (187). Notably, hypercidin is a potent inducer of immunogenic cell death (181).

The high affinity IL-2R (CD25) is overexpressed in the malignant T cells in CTCL (188). E7777 (formerly known as denileukin diftitox) is an engineered fusion protein combining IL-2 and the diphtheria toxin and binds to cells expressing the high affinity IL-2R, including Tregs and malignant T cells, causing endocytosis of the fusion protein (189). After endocytosis, ADP-ribosyltransferase activity of diphtheria toxin inhibits protein synthesis and causes cell death. Therefore, this therapy simultaneously decreases tumor burden and theoretically boosts the anti-tumor immune response by depleting immunosuppressive Tregs (190, 191). However, treatment has been shown to only target activated T cells, and thus resting Tregs remain in circulation (192). Furthermore, this treatment impairs dendritic cell maturation in vitro and in vivo, creating a tolerogenic dendritic cell phenotype, which induces T cell anergy instead of activated anti-tumor T cells (192). After being approved by the FDA in 1999, denileukin diftitox was taken off the market voluntarily in 2014 in order to create a more purified form with improved bioactivity to reduce the risk for systemic capillary leak syndrome (193, 194). E7777, a purified and reformulated denileukin diftitox, achieved a RR of 32% (six of 19 with PR and zero with CR) in patients with relapsed or refractory CTCL in a phase II trial (46). E7777 has also been investigated in a recently completed Phase III trial (clinicaltrials.gov identifier: NCT01871727), and application for FDA approval within the next year is anticipated after results are published. Resimmune, a similar fusion protein carrying cytotoxic diphtheria toxin, targets cells expressing the CD3ε surface molecule, which is found predominantly in T cells (195). In an inter-patient dose escalation trial in CTCL there was a 36% RR (nine out of 25) with four CRs (196). The modest response rates may be explained by its complex immune modulating function.

BNZ-1 an antagonist pegylated peptide that binds to the common gamma chain signaling receptor for certain cytokines, namely IL-2, IL-9, and IL-15, was investigated in MF/SS (45). It was hypothesized that inhibition of IL-2 and IL-15 would impair cytokine-driven tumor cell survival, inhibition of IL-2 and IL-9 would increase the anti-tumor immune response by decreasing Treg activity, and inhibition of IL-15 would create a beneficial anti-inflammatory effect (45). IL-15 is a putative autocrine survival factor found in elevated levels in MF skin samples and expressed by MF/SS cell lines (197). A phase I/II study of 19 patients with refractory MF/SS demonstrated positive results with 11 participants achieving PR and one CR; a phase 3 trial is being planned (45).

Mogamulizumab is a defucosylated, humanized, anti-CCR4 monoclonal antibody. CCR4 is the receptor for thymus- and activation-regulated chemokine (TARC/CCL17) and MDC/CCL22, a predominantly macrophage-derived chemokine (198). As mentioned, CCR4 is mainly expressed on CD4+ T cells (particularly memory T cells), CD4+CD25+FOXP3+ Treg cells, and only on a minority of CD8+ T cells (66). It is a putative marker for T cells with skin homing abilities, as it is thought to play a role in transendothelial diapedesis and epidermotropism (66). Not surprisingly, CCR4 is overexpressed in malignancies of skin homing T cells including MF, SS, and some peripheral T cell lymphomas (199). Mogamulizumab binds to the N terminal domain of CCR4, eliciting tumor cell killing via antibody-dependent cellular cytotoxicity (200). Defucosylation of the antibody’s Fc portion enhances binding-affinity of the Fc receptor on effector cells leading to more potent antibody dependent cellular cytotoxicity (ADCC).

In 2018, mogamulizumab was approved for relapsed or refractory MF/SS following a phase III clinical trial of 370 MF/SS patients, comparing mogamulizumab to vorinostat (50). Mogamulizumab achieved a global response of 35% compared to 6% in patients receiving vorinostat. The mogamulizumab group also had longer progression-free survival (median 7.7 months) compared to the vorinostat group (median 3.1 months). Infusion-related reactions, lymphopenia, drug rash [now termed mogamulizumab-associated rash (MAR)], diarrhea, and fatigue were the most common adverse events of any grade in patients receiving mogamulizumab (201). Immune-related adverse events, such as polymyositis, myocarditis, and hepatitis have also been observed, thought to be driven by Treg depletion (201). Recent retrospective cohorts have affirmed a high incidence of MAR in patients with CTCL (17 of 24, 68%) and found that it is associated with clinical responses, leading to the hypothesis that those with effective decreases in Tregs are more likely to experience MAR and achieve a clinical response (202). Patients who experience MAR and/or other immune related adverse events reportedly have higher rates of CR and PR as well as prolonged remissions of disease, in some cases over two years, giving credence to the theory that remodeling of the immune microenvironment contributes to short and long-term therapeutic efficacy (202, 203).

From the phase I/II study, mogamulizumab was observed to have a favorable effect on patients’ immune profile by reducing Treg cells and increasing NK cell killing function but no effect on the absolute number of NK and CD8+ T cell populations were noted at one month (204). However, a real world SS cohort found that mogamulizumab, in addition to depleting benign CD4+ T cells, Tregs, and malignant T cells, also decreased CD8+ T cells and NK cells at one month of treatment. Longer-term follow-up in this cohort revealed the emergence of “immune restoration” involving expansion of naïve and stem memory CD4+ subsets with almost complete disappearance of pre-treatment exhausted lymphocytes (PD-1+ or TIGIT+ CD4+ T cells) and activated Tregs (205). In skin biopsies taken from patients with MAR, an increase in interferon-stimulated gene IFI44L and macrophages expressing CXCL11 and CXCL9 coupled with expansion of new T cell clones expressing CXCR3 (the cognate receptor for CXCL9/11) and genes important for cytotoxic effector function was observed, shedding new insight on the potential role tumor associated macrophages may play in generating prolonged remissions from mogamulizumab (206). Despite favorable changes in the immune environment, disease relapse in the SS cohort appeared to be tied to the emergence of CCR4- malignant T cells (205).

Alemtuzumab, a monoclonal antibody, targets cells expressing CD52, which is expressed on T cells, B cells, monocytes, eosinophils, and a subset of dendritic cells (60). Alemtuzumab is primarily utilized for leukemic CTCL given its limited efficacy in clearing cutaneous disease. This is likely due to its primary mechanism of action, ADCC, which requires NK cells and neutrophils; these are present predominantly in the blood but not in the skin (207). Administration of alemtuzumab causes a depletion of lymphocytes, with depletion of Tregs and exhausted T cells being a favorable immune modulatory effect in CTCL (208). However, CD52 is broadly expressed in effector cells of the innate and adaptive immune response, important for anti-tumor immune responses and protection against infections (60). The overall RR in alemtuzumab is 55% with 32% of patients achieving a CR (47), though RR may be higher in select populations such as SS. Lower doses of alemtuzumab and anti-microbial prophylaxis have prevented many of the infectious complications seen in the earlier studies (47, 207, 209).

Lacutamab is a first-in-class humanized monoclonal antibody targeting inhibitory killer cell immunoglobulin-like receptor (KIR) KIR3DL2, which is aberrantly expressed in MF/SS and minor subpopulations of NK cells, CD8+ T Cells, and CD4+ T Cells (49, 210). In NK cells, KIR3DL2 is one of several inhibitory KIR receptors that recognizes MHC class I molecules (211, 212). Loss of MHC-I in tumor cells or decreased expression of inhibitory KIRs on NK cells leads to activation of NK cell-mediated killing. In MF/SS its function is less clear, but it appears to function as an inhibitory coreceptor—downmodulating CD3-dependent early signaling events following antigen recognition (213). Therefore, it is suspected that KIR3DL2 may protect malignant T cells from activation-induced cell death (AICD) (213). In pre-clinical studies of SS, lacutamab exerted an anti-tumor effect via ADCC, rather than through modulation of KIR signaling in malignant cells. A phase I clinical trial of 44 patients with relapsed or refractory cutaneous T cell lymphoma demonstrated 36% RR, predominantly in SS patients (49). Three patients experienced lymphopenia, which was the most common grade 3 or worse adverse event. The effect of lacutamab on NK cells in CTCL cohorts has not been evaluated. However, use of a related antibody directed against KIR2D inhibitory receptors (IPH2101) in multiple myeloma patients found unexpectedly that IPH2101 decreased the responsiveness of the NK cell. A phase II trial investigating lacutamab alone or in combination with chemotherapy in patients with advanced T cell lymphoma is underway (clinicaltrials.gov identifier: NCT03902184). Combination of lacutamab with PD-1 inhibition has been proposed, as stated above, given that SS responders to PD-1 inhibition had lower levels of KIR3DL2 (122).

Brentuximab vedotin (BV) is an antibody-drug conjugate, consisting of the chimeric monoclonal antibody to CD30 linked to monomethyl auristatin E (MMAE), a potent anti-tubulin agent. It was approved by the FDA for CD30+ cutaneous lymphomas in 2017 (214, 215). Upon binding to CD30, the antibody-drug conjugate is internalized, and MMAE is released causing G2/M arrest and apoptosis (216). For this reason, it is used to treat CD30+ lymphoproliferative diseases, including patients with large cell transformation of MF, which often express CD30 on tumor cells (217). Early clinical trials with 28 MF patients showed a RR of 54% with 7% achieving a CR (218). A median duration of response was 32 weeks. This response was not correlated with CD30 expression. Kim et al. evaluated the use of BV in MF with variable CD30 expression and found an overall response rate of 70% with 20 PR and one CR; they found that median CD30_max was higher in responders (13%) versus non-responders (3%) and least effective in cases with a CD30_max of <5% (219). This study suggested that BV can be effective in MF even with relatively low to intermediate levels of CD30 (CD30_max >5%). This may be attributable to the additional favorable immune modulatory effects, including depleting immunosuppressive M2 type tumor associated macrophages, inducing dendritic cell maturation, and promoting bystander cell death upon releasing MMAE (even among CD30- tumor cells) (214, 219). Further, BV was shown to deplete immunosuppressive inducible and primary CD30+ Tregs in a dose dependent manner in vitro and in vivo (132, 220). CD30+ CD8+ T cells, which play a role in anti-tumor immunity, however, were not affected by BV (132, 220). In a phase III clinical trial of 131 patients with CD30+ MF or primary cutaneous anaplastic large-cell lymphoma, BV was compared to physician’s choice of methotrexate or bexarotene (48). Among 48 MF patients, 50% achieved an objective global response lasting at least four months, outperforming the physician’s choice group.

The BCL-2 family consists of pro-apoptotic and pro-survival proteins. Shifts in balance between these two protein groups lead to cell death or survival (221). Overexpression of pro-survival BCL-2 proteins allows for various cellular stresses to occur that would otherwise promote apoptosis in non-cancerous cells, and overexpression of BCL-2 in cancers is associated with resistance to therapy (222–224). Likewise, BCL-2 expression is observed in CTCL lines that are sensitive to venetoclax, a BCL-2 inhibitor already approved by the FDA to treat chronic lymphocytic leukemia, small lymphocytic lymphoma, and acute myeloid lymphoma (225–227). An investigator-initiated, single-arm, open label study of venetoclax closed after enrollment of four patients (clinicaltrials.gov identifier: NCT04171791). A single case was published of a SS participant with skin and blood involvement treated with venetoclax who achieved a near CR (228). Despite no immediate plans for additional trials of venetoclax in CTCL, this agent may be best considered as part of a combination therapy to sensitize cells to apoptosis without impeding immune cell function (227). Control of apoptosis is an important mechanism regulating immune cell development and response (229). Despite decreases in T cell counts with venetoclax, largely due to reductions in naïve T cells, antigen-stimulated effector T cells can survive and proliferate via upregulation of other pro-survival proteins, such as BCL-xL (229, 230).

EZH2 is the enzymatic component of the polycomb repressive complex 2 (PRC2). PRC2 catalyzes trimethylation of lysine 27 of histone H3 (H3K27me3), which is deposited at CpG-dense promoters and is involved in gene transcriptional repression (231–233). EZH2 represses expression of tumor suppressor genes causing cell cycle progression and cell proliferation. Gain-of-function mutations in EZH2 have been associated with a variety of cancers, including T cell lymphoproliferative disorders (234–237). Additionally, there is a growing body of literature investigating the function EZH2 in the immune system, with many immune cells expressing EZH2 (238). Tumor infiltrating Tregs, for instance, appear to have EZH2 activity. Disruption of EZH2 activity in Tregs promotes pro-inflammatory functions and increases recruitment and function of CD8+ and CD4+ effector T cells (239). Inhibition of EZH2 in NK cells improves NK cell function (240). However, inhibition of EZH2 can impair effector T cell function and dendritic cell antigen presentation (240, 241). Currently, valemetostat (DS-3201b) a highly specific inhibition of EZH2 and EZH1 is being evaluated in a phase II trial for patients with relapsed or refractory peripheral T cell lymphoma, including non-MF/SS CTCL (NCT04703192). A phase I study of 15 patients with relapsed or refractory non-Hodgkin lymphomas, including PTCL, treated with valemetostat had a 53% RR (seven PR, one CR) (242, 243). Among the five patients with T cell lymphoma the RR was 80% (three PR, one CR).

There are 18 histone deacetylases (HDACs), which are categorized into four major classes: class I (HDAC1, 2, 3 and 8), class II (HDAC4, 5, 6, 7, 9, and 10), class III (aka sirtuins; SIRT1, 2, 3, 4, 5, 6, and 7), and class IV (HDAC11) (244). In general, histone deacetylation induces gene silencing. The clinical utility of HDAC inhibitors (HDACi) may include disruption of the epigenetic balance in malignant cells or activation of tumor suppressors, but other “off-target” effects include acetylation of transcription factors modulating their function. The current HDACi that are approved or under development are “pan-HDAC” inhibitors (inhibit HDACs in Class I, II, or IV), class specific inhibitors for either class I or class II, or inhibitors of a single HDAC protein. For example, vorinostat is a pan-HDACi and romidepsin inhibits class I HDACs. In multiple cancer models HDACi have been shown to affect several pathways, which favorably lead to cell cycle arrest, induction of apoptosis, and inhibition of angiogenesis (245). However, HDACi have pleiotropic immune modulating functions, which may have opposing effects on the anti-tumor response.

The “off-target” effects of HDACs on the immune system are diverse with both desirable and undesirable results. HDACs can have immunosuppressive effects. Several HDACs in class I, II, and III have been shown to inhibit TLR mediated expression of pro-inflammatory genes such as the p40 subunit of IL-12, IFN-β, and NF-κB activation (244). Class I HDACs have also been shown to inhibit MHC II expression, thereby decreasing antigen presentation. HDACs also increase PD-1L expression (246).

HDACs also have a multitude of immune activating functions. Class I HDACs have been shown to be important for activation of STAT1 and STAT2 dependent interferon signaling (247). Class IIa HDACs, notably HDAC7, have been shown to positively regulate TLR mediated expression of pro-inflammatory cytokines in macrophages, cytokines, and associated receptors for cytotoxic T lymphocyte function (248, 249). Class IV HDACs are thought to play a role in promoting antigen presentation and antigen specific T cell responses (250). Therefore, inhibition of these pathways could impair the interferon-mediated immune response to viral, bacterial, and fungal pathogens and have a negative effect on the adaptive immune response. In fact, mice treated with HDACi are more susceptible to bacterial and fungal pathogens (251).

Given the overlapping and opposing actions of HDACs on the immune system it is unclear if selective inhibition of a particular class of HDACs or an individual HDAC protein would be more beneficial in CTCL. HDACi have been shown to increase tumor immunogenicity, have both a positive and negative regulation of the innate and adaptive immune system via modulation of TLR and interferon signaling pathways, and divergent effects on Treg function (244). Inhibition of class I HDACs has been shown to decrease Treg populations and increase immunologic cell death via upregulation of FAS, FASL and TRAIL. However, pan-HDACi, likely due to inhibition of class II HDACs, have been shown to increase Tregs and their function (252). Resistance may be related to an inability to orchestrate a favorable TME or may be related to properties intrinsic to the tumor cell. At the single cell level, for instance, SS has been shown to have a high degree of heterogeneity with differing sensitivity and resistance to HDACi (253). Clinical trials of romidepsin have shown an overall RR of approximately 30% with 6% of patients achieving CRs (41). As suggested above, infections (58% of patients for romidepsin) were common adverse events (254). Vorinostat and romidepsin have been shown to suppress NK and dendritic cell function in patients with CTCL (255). In a patient with SS that effect could be overcome with addition of an immune activating agent, such as TLR agonists or IFN-α when evaluated in vitro (255, 256). Therefore, future rational combination therapies with an immune activating agent and HDACi may synergistically enhance the anti-tumor response and decrease the risk of infections. Vorinostat and romidepsin are both approved for use in CTCL. Additional HDACi have been investigated in CTCL including panobinostat with 20% RR (40), belinostat with 14% RR (257), topical SHAPE gel also known as SHP-141 with interim results of 32% showing mSWAT response at 6 months (258), and resminostat with no published results (259).

Lenalidomide is a derivative of thalidomide (260). It binds E3 ligase protein cereblon (CRBN) directing protein ubiquitination and degradation of the IL-2 transcriptional repressors Ikaros (IKZF1) and Aiolos (IKZF3) (260, 261). Through modulation of these transcription factors, and likely others, it is able to have diverse effects on the innate and adaptive immune system. These effects include an increase in antigen presentation capacity, expansion and increased activity of cytotoxic CD8+ T cells, and a decrease in T_h2 cytokine expression (261). In a phase II study of lenalidomide in refractory MF/SS, there was an overall RR of 28%; all were PR. The median duration of response was 10 months (39). In these patients, while there was an overall decrease in malignant CD4+ T cells and an increase in CD8+ T cells in MF skin lesions, an increase in CD25+ and FOXP3+ staining was also observed (39). This increased T regulatory phenotype may be secondary to the increase in IL-2 induced by lenalidomide.

The JAK-STAT pathway, which transduces signals from cytokines, interleukins and growth factors, has been shown to be dysregulated in both MF and SS (4, 262). The signaling begins with binding to the cell-surface receptor which results in dimerization and causes activation of JAK tyrosine kinases. Subsequently, activated JAKs phosphorylate specific tyrosine residues on the receptor and act as docking sites for STATs, which are cytoplasmic transcription factors (61). Once STAT proteins bind to the receptor, the JAKs phosphorylate the STAT proteins, which dimerize and translocate to the nucleus to induce expression of genes involved in inflammation. STAT4, which is required for T_h1 differentiation and is activated downstream of IL-12 signaling, has decreased expression in the skin and blood of patients with advancing CTCL stage (262). STAT5, activated downstream by IL-2, IL-7 and IL-15 signaling, and STAT3, activated downstream by IL-2, IL-6, IL-7, IL-9, IL-10, IL-15 and IL-21, may be constitutively activated independent of cytokines in MF/SS (262–264), although there is conflicting data (265). Constitutively activated STAT3 can increase survival and resistance to apoptosis in malignant T cells by increasing the anti-apoptotic protein BCL-2 (266). Furthermore, activation of STAT3 has been shown to induce expression of immunosuppressive ligand PD-L1 (77). STAT5 induces anti-apoptotic proteins, cell cycle genes (cyclin D and c-myc) and IL-4, which promotes carcinogenesis (61). In addition, its activation induces microRNA-155 that degrades STAT4 further solidifying the T_h2 bias in the microenvironment (262, 267, 268). Furthermore, STAT3 and 5 are also important for the function of Tregs (269). Interestingly, methotrexate has been found, using a luciferase-based transcriptional assay in a Drosophila model system where a library of 2000 small molecules were screened, to be a potent inhibitor of the JAK-STAT pathway at doses that are equivalent to that seen in patients taking low-dose oral methotrexate (113).

While inhibition of JAK appears to be an attractive therapeutic candidate, malignant T cells do not exist in an isolated environment. As mentioned in the biologic response modifiers section, type I and II interferons activate the JAK-STAT pathway, leading to expression of interferon response genes with anti-proliferative and pro-apoptotic effects on malignant T cells. It is possible that the favorable effects of interferon-associated JAK-STAT signaling are blunted by JAK inhibitors. TLR-stimulated plasmacytoid dendritic cells, known for their ability to produce high levels of IFN-α for instance, had robust inhibition of IFN-α production in response to JAK inhibitor tofacitinib in vitro (270). Furthermore, with constitutive activation of STATs downstream from JAK in some patients, there is controversy over the ability of JAK inhibition to overcome downstream activation.

The FDA approved JAK inhibitors, including tofacitinib and ruxolitinib, demonstrated anti-tumor properties in CTCL cell lines (271–273), which has subsequently led to recent clinical trials. In a phase II trial of patients with relapsed or refractory PTCL or CTCL receiving ruxolitinib, a 20% RR was achieved among the five evaluable patients with CTCL (42). A phase IIa study of 37 patients with CTCL and 61 patients with PTCL investigated a small molecule spleen tyrosine kinase (SYK)/JAK kinase inhibitor called cerdulatinib (274). SYK is a tyrosine kinase, with its signaling pathway thought to be an oncogenic driver in multiple cancers, including immature hematopoietic neoplasms with T cell differentiation (275). The RR was 35% in CTCL patients overall. In MF the RR was 45%, with 9% of these participants achieving CR (37). Patients with Sézary syndrome had a 17% RR with no CRs.

The PI3K-Akt pathway has been implicated in several cancers and is an important signal transduction pathway for immune cells (276, 277). There are three classes (I-III) of PI3Ks (278). Class I, the most relevant to oncology, consists of four isoforms: α, β, δ, and γ. (271) PI3K γ and PI3Kδ are expressed in hematopoietic cells, but PI3Kδ is the dominant isoform in T cells, mediating downstream signaling of the TCR, costimulatory and cytokine receptors (279). PI3Ks phosphorylate the 3-hydroxyl group of the inositol ring found in phosphatidylinositol lipid substrates (280). The 3-phosphoinositides (PIP₃) mediate the function of multiple effector proteins that bind these lipids. In turn, PI3Ks play a role in intracellular vesicular transport, cell cycle progression, and cell growth, survival, and migration.

Activating mutations in the PI3K-Akt pathway and constitutive Akt activity have been observed in CTCL, supporting the importance of PI3K signaling for malignant T cell survival and proliferation (38). Duvelisib, a PI3Kγ and PI3Kδ inhibitor, was approved by the FDA in 2018 for refractory chronic lymphocytic leukemia and small lymphocytic leukemia and has demonstrated activity in T cell lymphoma cell lines (281, 282). In a phase 1 study of duvelisib for patients with relapsed or refractory PTCL or CTCL, a RR of 31.6% (six of 19) was observed in the CTCL cohort (38). Of note, PI3K inhibition may also affect the immune microenvironment. In PTCL patient-derived xenografts treated with duvelisib a shift in polarization from immunosuppressive M2 macrophages to an inflammatory M1 phenotype, which promotes anti-tumor responses, was observed (38). In other murine models, P13K inhibition of tumor associated myeloid cells was able to sensitize tumors to ICI (131). However, these murine models lack an intact immune system to understand the full effects of P13K inhibition on the microenvironment. For example, PI3K signaling is important for antigen- and chemokine-dependent effector cell trafficking to peripheral sites of inflammation through regulation of leukocyte function-associated antigen-1 (LFA-1) (279). LFA-1 is needed for transendothelial egress and establishment of immunological synapse with antigen presenting cells (279). Therefore, PI3K inhibition may limit migration of anti-tumor effector T cells to the skin and effective engagement with antigen presenting cells thereby hindering the anti-tumor response. However, treatment of CD8+ T cells ex vivo with duvelisib, leads to enhanced T cell activation and cytotoxicity (283). Therefore, the net result of PI3K inhibition is unclear. Tenalisib (RP6530), another dual PI3Kδ and PI3Kγ inhibitor, is also under investigation, with a recent phase I/Ib study of relapsed or refractory PTCL and CTCL and RR of 45% was observed (43). A phase I/II study evaluating tenalisib in combination with romidepsin was recently completed (NCT03770000). Duvelisib was also investigated in combination with romidepsin or bortezomib in a phase I/II trial of PTCL/CTCL, based on in vitro drug screens predicting synergy. The RR in CTCL was 46% with romidepsin and 28% with bortezomib; no CRs were observed in either group (284).

Proteasomes are fundamental to the ATP-dependent proteolytic pathway in eukaryotic cells (285). In fact, most cellular protein degradation occurs through proteasome catalyzed degradation (286). The FDA approved the first proteasome inhibitor, bortezomib (PS-341), for multiple myeloma in 2003 (287, 288). Bortezomib is a modified dipeptidyl boronic acid that is a reversible inhibitor of the 26S proteasome specifically (287). Bortezomib has pro-apoptotic and anti-proliferative effects via inhibition of NF-kB and induction of the unfolded protein response (289). It has also been shown to increase tumor cell killing by dendritic cells, likely due to induction of immunogenic cell death (290), and by NK cells (291, 292). A phase II trial of 15 patients with MF and unspecified peripheral T cell lymphoma explored the anti-tumor properties of bortezomib (28). Among the 10 patients with MF the RR was 70%, with one CR and six PRs. Responses were remarkably durable, with the CR in remission at the latest follow up at 12 months and with the PRs lasting at least seven months.

Several retinoids have been used in the treatment of CTCL, including etretinate, acitretin, all trans retinoic acid (ATRA), alitretinoin and isotretinoin (293). Retinoids exert their biologic effect via binding and activation of the retinoic acid receptor (RAR) isoforms RAR α, β, and γ, which form a heterodimer with each other or with retinoid X receptors (RXRs), or, in the case of isotretinoin, through indirect effects on the RAR signaling pathway (293). They have anti-proliferative, pro-apoptotic effects in CTCL, as well as immune modulating effects (294). Retinoids can induce IFN-γ partly via IL-12 (295), upregulate Langerhans cell antigen presentation capacity and coreceptors required for T cell activation (296), and increase NK cell activity (60).

Bexarotene, a novel rexinoid, is a ligand for the RXR isoforms: RXR α, β, and γ. RXR can form heterodimers with the RAR, vitamin D receptor, thyroid hormone receptor, peroxisome proliferator activator receptor, liver X receptor, and the farnesoid X receptor to induce gene transcription. Bexarotene has anti-proliferative and pro-apoptotic effects on malignant cells (60). It has been shown to downregulate T_h2 cytokines (297) and limit migration of malignant T cells via decreased expression of CCR4 on T cells and decrease of E-selectin on endothelial cells (298). Additionally, bexarotene in MF patients reduces CCL22 production by M2-polarized tumor associated macrophages (299). CCL22, similar to CCL17, can recruit Tregs (65). Bexarotene was also shown in leukemic cell lines to increase expression of IL-2R, which may be beneficial if combined with E7777, which targets IL-2R expressing cells (discussed above) (300). Bexarotene induced an objective RR of 45% with 13% of patients achieving a CR (27). Bexarotene has been used in combination with many other treatments, including PUVA, interferons, ECP, and others, and while this combination therapy appears safe it is unclear if there is appreciable benefit (60, 301).