Introduction

Cutaneous T-cell lymphomas (CTCL) are a group of rare lymphomas originating from mature skin-homing T-cells [1, 2]. Mycosis fungoides (MF) and Sézary syndrome (SS) are the most common variants of CTCL [3]. Though they historically were considered distinct diseases by the World Health Organization and the European Organization for Research and Treatment of Cancer based on their differing presentations, others have posited that Sézary Syndrome, is the result of disease progression from earlier stages [4]. We have shown that underlying this difference in clinical presentation is likely a distinct cell of origin: skin resident memory T-cells (TRM) in MF and skin-tropic central memory T-cells (TCM) in SS [5,6,7]. These T-cell subsets are defined by the expression of cell surface molecules, which dictate where and how often they recirculate from the skin, blood, and lymph node. TRM cells typically to not recirculate out of skin, and clinically are associated with fixed and discrete lesions, e.g. plaques, typical of MF. TCM cells recirculate between the skin, blood, and lymph node, and clinically are associated with diffuse erythema, as in SS, where patients commonly present with erythroderma. However, leukemic disease may be occur in the presence or absence of malignant TRM like cells.

Alemtuzumab, a humanized monoclonal antibody targeting surface antigen CD52, was approved by the Food and Drug Administration in 2001 for the treatment of chronic lymphocytic leukemia [8] and produces high response rates in patients with chronic lymphoproliferative disorders [9]. Additional approval was obtained in 2014 for the treatment of multiple sclerosis [10]. Alemtuzumab, in both high (30 mg three times a week, intravenously administered) and low doses (10 mg three times a week, subcutaneously administered), has been used off label for the treatment of advanced-stage leukemic CTCL (L-CTCL). Alemtuzumab depletes all cells expressing CD52 from blood but does not deplete cells in tissues and thus spares non-migratory TRM in the skin [11]. In early phase clinical studies, low-dose alemtuzumab (LDA) was shown to produce robust responses in patients with erythrodermic CTCL [12, 13]. LDA is preferred in CTCL given increasing risk of infection at higher doses [14, 15].

Given the dearth of clinical trials using LDA in CTCL, primarily due to the rarity of the disease, our goal was to characterize patients with CTCL who may benefit the most from LDA treatment using real world data. Patients with CTCL presenting with erythema with no prior or present history of papules, plaques, and tumors (PPT) have a malignant T-cell phenotype associated with TCM. In this study, we evaluate whether this presentation of CTCL predicts better global complete response (CR) and skin responses compared to patients having a history of PPT, the latter associated with a TRM phenotype. We also evaluated time to next systemic treatment, time to cessation of pruritus, rates of infection, and overall survival (OS) with 5 years’ follow-up.

While we report that the initial clinical presentation of CTCL (SS vs MF) is likely reflective of the cancer cell of origin (TCM vs TRM), there is also likely a continuum between the extremes of these presentations [7]. Accordingly, malignant T-cells may exhibit some phenotypic plasticity and thus the clinical morphology of disease in individual patients, rather than a clinical diagnosis of MF vs SS, may be a more effective strategy to identify who will benefit from alemtuzumab. We further hypothesized that in patients with CTCL, the best responders will be those with a pure TCM clinical phenotype, defined by an absence of PPT and presenting with diffuse erythema [11].

Methods

Study design, setting, and participants

We performed an institutional review board approved retrospective cohort study, reviewing the medical records of 40 patients with CTCL referred to the Dana-Farber/Harvard Cancer Center and treated with LDA, 10 mg of subcutaneous alemtuzumab three times weekly, between January, 2002 to December, 2014 with at least 5 years of follow up. Two patients were excluded because they were seen in consultation only.

Data collection

We abstracted patient demographics, clinical features and labs. History of PPT was defined as history or presence of cutaneous lesions characterized by palpable papules, plaques, or tumors. We classified patients with CTCL within 3 clinical morphologies based on the predicted immunophenotype of their malignant T-cells: (1) TCM, (2) TRM, and (3) migratory memory T-cell (TMM) [6, 11] TCM patients had erythroderma, defined as >80% body surface area (BSA) with erythema, and no history of PPT. Patients with TRM phenotype had PPT and no diffuse erythema. Patients with a TMM phenotype had a history of or active PPT and diffuse erythema. We classified response to LDA treatment as CR, partial response (PR), stable disease (SD), or progressive disease (PD) based on response criteria as defined by Olsen et al. except skin assessment was based on the percentage change in total BSA affected [16].

Neutralizing antibody detection

Alemtuzumab was labeled with Alexa Fluor 488 using the Alexa Fluor 488 Microscale Protein Labeling Kit (Life Technologies, A40006). Healthy donor peripheral blood mononuclear cells (106 cells/test) were mixed with FACS buffer (PBS, 2% FBS, 0.1% sodium azide), anti-CD3-PerCP (1:100; Biolegend # 300428), alemtuzumab-Alexa Fluor 488 (1:200), and indicated concentrations of healthy donor or patient plasma for 30 m at ambient temperature. Cells were washed twice and fixed with 0.5% paraformaldehyde. Cells were analyzed on a Becton Dickinson FACSCanto and data were analyzed using FACSDiva software (V8.0.1).

Statistical analyses

Summaries of demographics, disease and prior treatment for patients are based on descriptive statistics. Comparisons of patient characteristics according to history of PPT were conducted using Wilcoxon rank-sum tests for continuous measures and Fisher’s exact tests for categorical measures. Response rates are summarized using proportions with 95% exact binomial confidence intervals. Analyses of response according to PPT were based on multivariable logistic regression models. Time-to-event outcomes were modeled using proportional hazards regression in the presence of the competing risk of transplant. Comparisons of cumulative incidence functions are based on Gray [17] and proportional hazards regression models are based on Fine and Gray [18] using candidate predictor variables. CD4 and CD8 count, percentage of eosinophils, lactate dehydrogenase (LDH), and white blood cell count (WBC) at first dose of LDA were log2 transformed to improve linearity. For logistic and Cox regression models, both backward elimination and stepwise procedures were used to insure model consistency. Statistical significance in the modeling was based on the Wald’s test and used ≤0.05 as a decision point. Analyses were conducted using SAS 9.4 (SAS Institute Inc., Cary, NC, USA).

Results

Patient demographics and characteristics according to PPT

Thirty-eight patients with CTCL were treated with LDA. None had large cell transformation at the time of LDA treatment. Eleven (29%) patients had no history of PPT (TCM), and 27 (71%) patients had a history of PPT (TRM/TMM). Patient clinical and histologic characteristics are summarized according to history of PPT, shown in Table 1. Univariate comparisons were performed considering the following candidate variables: age at first LDA (mean 66.8 years, range 45.0–85.0), CD4/CD8 ratio at first LDA (24.6, 1.5–96.0), CD4 count at first LDA (4307.2 cells/uL, 411.0–27,771), WBC at first LDA (12.5 × 109 cells/L, 5.1–54.3), LDH at first LDA (278.2, 113.0–989.0), percentage of eosinophils at first LDA (5.2%, 0.0–43.0%), peak LDH (332.3, 113.0–1332.0), peak % of eosinophils (7.0%, 0.0–43.0%), number of prior therapies (3.1, 0.0–11.0), and years between diagnosis and treatment (3.0, 0.1–16.1). Univariate comparisons showed differences according to PPT with respect to peak % of eosinophils (Wilcoxon rank-sum p < 0.001). Lesser relationships (p < 0.1) were observed for age > 60 years (Fisher’s exact p = 0.08) and T-stage (Fisher’s exact p = 0.06).

Table 1 Patient characteristics according to history of papules, plaques or tumors (PPT).
Full size table

Global response and complete response rates according to PPT

Responses according to T-cell phenotype are summarized in Table 2. The overall response rate to LDA was 63.2% (95% exact CI: 46–78%). Complete responses were achieved by 44.7% of patients (95% exact CI: 29–62%). All patients with TCM phenotype experienced a CR or PR to LDA therapy compared to 53.5% with a TRM or TMM phenotype. Five (13.2%) patients did not respond to LDA therapy (TRM = 2, TMM = 3); all had active PPT at the time of treatment. Patients with TCM phenotype had a superior CR rate of 72.7% (95% exact CI: 39–94%) compared to TRM or TMM phenotype [CR = 33% (95% exact CI: 16–54%), p = 0.03]. Among patients with leukemic disease (i.e. B2 disease, n = 29), the CR rate was 96.5% (CR = 28, PR = 1) in the blood compartment.

Table 2 Global response according to history of papules, plaques or tumors (PPT).
Full size table

Patients with a history of PPT (n = 27) were further disaggregated into those with and without active PPT at time of LDA treatment. Sixteen patients had active PPT at time of LDA treatment with a CR rate of 18.7%. Eleven patients did not have active PPT at time of LDA treatment and had a CR rate of 54.6%.

Multivariable prediction of CR

Based on univariate analyses, PPT, years between diagnosis and treatment, age >60, LDH at first treatment, and percentage of eosinophils at first treatment were evaluated as candidate predictors of CR. PPT (Pr > Chi Sq = 0.03), LDH at time of first LDA dose (Pr > Chi Sq = 0.047), and years between diagnosis and treatment (divided at the median of 1.5 years) (Pr > Chi Sq = 0.04) were found to be significant predictors of CR. Each 10-unit increase in LDH at the time of first treatment decreased the likelihood of CR [OR: 0.9, 95% CI (0.8–0.99)]. Patients with less than 1.5 years between diagnosis and treatment were less likely to experience CR [OR: 0.17, 95% CI (0.03–0.95)]. When controlling for these covariates in the logistic regression, patients with no history of PPT (TCM) were more than 8 times as likely than patients with history of PPT (TRM/TMM) to achieve CR [OR: 8.2, 95% CI (1.2–57.6)]. The c-index for this model is 0.85, indicating very good ability to classify patients according to presence or absence of CR.

Skin response according to history of PPT and active PPT at time of treatment

Skin response according to PPT categories are summarized in Table 3. Among 38 patients, 19 (50%) had a CR in the skin. The skin CR rate was 81.8% in patients with TCM compared to 37.0% in patients with TRM/TMM (OR: 7.7, 95% CI (1.4–42.7)). For patients with a history of PPT, the skin CR rate was 82.4% in patients without active PPT compared to 17.7% in patients with active PPT at time of LDA (OR:7.6, 95% CI: 1.6–34.9).

Table 3 Skin response according to history of papules, plaques or tumors.
Full size table

Outcomes of CTCL patients treated with LDA therapy

5-year survival outcomes for L-CTCL patients treated with LDA

The median follow-up time for patients who were alive or underwent transplantation was 5.3 years (interquartile range [IQR]: 0.7–9.4 years). The estimated cumulative incidence of death was calculated to allow for the competing risk of transplant (Table 4). Fifty percent of patients died by 5.2 years (95% CI: 33–65%). The estimated incidence of death at 2 years was 24% (95% CI: 12–38%), and at 5 years was 47% (95% CI: 31–62%) (Fig. 1). When evaluating the cumulative incidence of death according to TCM/TRM/TMM timed from the first dose of LDA, no significant difference in the incidence or cause of death (Fisher’s p = 0.58) between patients with and without history of PPT was found (Table 5). Among patients with no history of PPT, 27% died by 2 years (95% CI (6–55%)); 22% of patients with a history of PPT died by 2 years (95% CI (9–39%)). Based on the Fine and Gray regression model using transplant as a competing risk, PPT did not predict survival after the first LDA dose (univariate HR for PPT no vs. yes: 0.93, 95% CI (0.36–2.40); p = 0.88).

Table 4 Outcome events according to PPT.
Full size table
Fig. 1: Cumulative incidence of death according to PPT.
figure 1

The estimated incidence of death at 2 years was 24% and at 5 years was 47%, here illustrating no significant difference noted in survival in those with and without a history of PPT.

Full size image
Table 5 Cause of death according to history of papules, plaques or tumors (PPT).
Full size table

Time to next systemic treatment

Among patients with TCM phenotype, 22% required and intiated new, non-transplant treatment at 12 months (95% CI: 3–53%), 40% of patients with TRM/TMM phenotypes initiated new, non-transplant treatment at 12 months (95% CI: 20–59%). There were no significant predictors of time to subsequent systemic treatment; PPT was not a significant predictor of time to subsequent treatment after the first LDA dose [HR (PPT No vs. Yes): 0.57, 95% CI (0.3–1.4); p = 0.22]. Given excellent responses observed in the blood, sixteen patients (42%) were treated with a second course of LDA, four patients received a third course, and one patient received a fourth course. Median interval between courses of LDA was 29.5 months, range 10–82 months. Ten patients had persistent or progressive disease on LDA during initial or subsequent courses. Seven had complete or significant loss of CD52 by peripheral blood flow cytometry of the malignant clone (range, 0–22% CD52 positive); two retained CD52 expression (79% and 100%). Among the latter, alemtuzumab neutralizing antibodies were detected in one patient (Fig. 2).

Fig. 2: Alemtuzumab neutralizing antibody assay with plasma of patient with endogenous neutralizing anti-alemtuzumab antibodies.
figure 2

Flow cytometry was performed on healthy donor peripheral blood mononuclear cells (PBMC) incubated with Alexa Fluor 488-labeled alemtuzumab in the presence of healthy donor (no neutralizing antibodies) or CTCL patient plasma. Top panels: In the absence of plasma or with healthy donor plasma, >98% of CD3+ cells bind fluorescently labeled alemtuzumab. In the presence of positive control plasma with known neutralizing anti-alemtuzumab antibodies, 96.7% of CD3+ cells remain unbound by labeled alemtuzumab. Bottom panels: In the presence of increasing concentrations of test patient plasma (10–50%), CD3+ cells display decreasing binding of labeled alemtuzumab, indicating increasing concentration of anti-alemtuzumab neutralizing antibodies in the patient plasma.

Full size image

Time to cessation of itching

Among TCM phenotype patients, 54.5% were free from itching at 1 month [95% CI (20–79%)] and all reported cessation of itching by 3.4 months. Twenty-six percent of TRM/TMM patients were free from itching at 1 month [95% CI (11–43%)], 70.4% by 3.4 months [95% CI (48–84%)], and 77.8% [95% CI (55–90%)] by 12 months. History of PPT and CD4:CD8 ratio were significant predictors of cessation of itching. When adjusted for CD4:CD8 ratio at first treatment, patients with no history of PPT were significantly more likely to cease itching than patients with a history of PPT [HR (PPT no vs. yes): 2.42, 95% CI (1.2–4.8); p = 0.01]. In addition, patients with higher CD4:CD8 at the time of first LDA were less likely to stop itching; a doubling of CD4:CD8 resulted in a 20% reduction in the hazard of itch cessation (HR (doubling of CD4:CD8): 0.80, 95% CI (0.64–0.99); p = 0.04).

Infections in L-CTCL patients while on LDA therapy

Eight (24%) patients experienced infectious complications on LDA including dermatomal varicella zoster virus (non-adherent to prophylaxis), perirectal human papillomavirus recurrence, a mild upper respiratory tract infection, tinea corporis, and increase in cytomegalovirus titer. Three (8%) patients experienced serious infections including aspergillus pneumonia in a patient with a history of chest radiation for Hodgkins disease, methicillin-sensitive Staphylococcus aureus (MSSA) bacteremia in the setting of poor access to hygiene measures secondary to homelessness, and pneumonia in a patient with congestive heart failure, dementia, and recent hip fracture requiring systemic intervention and hospitalization.

Discussion

The management of CTCL is largely driven by clinical stage, which is a major determinant of risk of disease progression (RDP) and overall survival (OS) [5]. There is a paucity of effective therapies for patients with advanced-stage disease [5]. Alemtuzumab has been increasingly used in the treatment of L-CTCL, but clinical trials are few, though the drug was first introduced for hematologic malignancy in 2001 [19]. Alemtuzumab is a humanized monoclonal antibody that targets CD52, a surface antigen expressed in high quantities on all mononuclear leukocytes in blood, including malignant T-cells, and induces antibody-dependent cellular cytotoxicity (ADCC) as a predominant mechanism of cell killing [19, 20]. The mechanism of T-cell depletion by alemtuzumab requires the presence of neutrophils, a key effector cell that mediates ADCC [21, 22]. Neutrophils are present in the blood but are rare in normal skin, which may explain decreased efficacy in clearing non-migratory TRM cells in the skin [11]. Our cohort supports the use of LDA for CTCL with blood involvement and further describes the patient population most likely to benefit. Choosing a therapy based upon clinical characteristics of the circulatory phenotype of the malignant cells has not been previously reported.

In this study, we evaluated a cohort of 38 patients with CTCL treated with LDA with a minimum of 5-years of follow-up data. We separated our patients into two groups based upon differing clinical phenotypes: those with no history of palpable disease (never PPT/ TCM phenotype) (n = 11), and those with history of PPT (n = 27) (TRM or TMM phenotype). Given the limited efficacy of alemtuzumab in MF patients (non-migratory TRM phenotype), we predicted that among patients treated with LDA, the best responders would be those with the clinical phenotype associated with TCM. We do not expect the malignant cells in PPT (TRM phenotype) to recirculate; therefore they are protected from ADCC.

In our cohort, those with an exclusively TCM phenotype had superior outcomes (best global response, CR rate, skin response, time to cessation of itch) and no difference in overall survival. This was true despite this group being predominantly male (68.4%), older, and having T4 stage disease, all risk factors for poorer prognosis [23]. Fully 100% of patients with TCM phenotype responded with either CR or PR to LDA therapy, with CR rate of 72.7%. This was significantly better than the CR rate of 33% in patients with a history of PPT. Patients with the TCM phenotype had 8 times the odds of patients with the TRM/MM phenotype to achieve CR on LDA therapy in the multivariable logistic regression model. When analyzing the data further among patients with history of PPT, we found that those with history of PPT - but no active PPT - responded better to LDA than those with active PPT at time of LDA. In summary, our findings suggest that LDA monotherapy can produce exceptional response rates in terms of global and skin response in CTCL with the pure TCM phenotype and those with no active PPT at the time of treatment (presumably lower TRM burden). Due to inferior responses of LDA in the skin of patients with PPT, the addition of skin-directed therapies is strongly recommended in combination with LDA. The presence of PPT did not impact responses to LDA in the blood. We found that baseline elevated LDH, an indirect measure of disease burden, was associated with a lower likelihood of achieving a global CR, possibly driven by those with greater, bulkier disease burden in the skin.

Our study’s findings align with the current understanding of CTCL and alemtuzumab’s mechanism of action. There is compelling evidence that SS and MF arise from distinct skin homing memory T-cell subsets: TRM in MF and TCM in SS [5,6,7]. This difference in T-cell origin can explain their unique clinical presentations. In SS, malignant TCM cells are migratory traveling from skin to blood via lymphatics; hence, patients present with diffuse blanching erythema [7]. In MF, malignant TRM are predominantly non-migratory, causing fixed PPT in the skin [7]. However, there can be significant clinical and histopathological overlap among MF and SS patients, which may suggest an inherent plasticity in malignant T-cell migratory phenotype [7]. Patients with CTCL who present with the TCM phenotype include those with SS and less commonly with erythrodermic MF [24] Patients with leukemic disease may sometimes have fixed PPT with or without a background of diffuse erythema (i.e., leukemic MF). Therefore, we believe the clinical morphology (pure TCM vs TRM/MM) at the time of treatment initiation is a more useful predictor of clinical response than stage or disease classification, as LDA works by depleting all migratory T-cells (e.g. TCM) but spares TRM. It is not surprising then that mogamulizumab, an anti-CCR4 antibody that works via ADCC, also has superior responses in the blood compared to the skin and is more effective for SS than MF [25].

Our findings are consistent with observations made in prior studies by us and others showing the superiority of alemtuzumab in patients with SS over MF [6, 14]. In a recent systematic review (n = 308; 93 with SS and 147 with MF), alemtuzumab was found to be significantly more effective in patients with SS [ORR: 81%, CR rate (CRR): 38%] than patients with MF (ORR: 29%; CRR: 8%) [15]. Furthermore, within the phase II study of dose escalating alemtuzumab (up to 30 mg intravenously), the presence of erythroderma had higher responses (ORR 69%) than those with plaques or skin tumors (ORR 40%), but did not investigate responses across the full spectrum of disease presentations, e.g. considering those with TMM phenotype (fixed plaques with intervening background erythema). Taken together, this study adds to available data to define the clinical presentations most likely to respond to LDA and identifies those patients in whom combination skin directed therapy with LDA would be most appropriate.

We also note improved 5-year overall survival (OS) in patients treated with LDA compared to known OS in cohorts of patients with advanced-stage CTCL published in the literature before the commercial availability of alemtuzumab [26]. Most patients in our cohort had stage IV disease (n = 32, 84.2%). At 5 years, the cumulative incidence of death was 47% (95% CI: 31–62%). A crude incidence of death, based on prior reported 5-year OS rates range from 60 to 85% for stage IVA [23, 27]. The lower incidence of death for patients treated with LDA is remarkable compared to the low survival outcomes in patients with stage IV CTCL in these published cohorts. Our results are consistent with a recent meta-analysis of 10 studies, all but two published after 2010, reporting a 5-year OS and defined “best case” OS reflecting the upper 10th percentile for each stage and “worst case” as the lower 10th percentile for each stage [28]. The crude incidence of death, based on five-year OS, is 47.5% in Stage IVA1 (best 39%, worst 56.8%), 66% in stage IVA2 (best 53.7%, worst 77.9%), and 76.7% in Stage IVB (best 60.3, worst 90%). In our study, there was no significant difference in the incidence of death between patients with and without a history of PPT. This shows that while LDA may produce a robust global and skin response in patients with the TCM phenotype, overall survival is likely most dependent on responses in the blood, not skin. Lastly, patients who relapse following treatment with LDA (typically given as a 12-week course), may successfully be retreated with LDA, provided expression of CD52 is not lost by the malignant clone and neutralizing antibodies do not develop.

In patients with advanced-stage CTCL, infections are a source of concern and a common cause of morbidity and mortality due to disrupted skin barrier and decreased cell-mediated immunity [29, 30]. Eight patients in our study experienced infectious complications on LDA treatment (24%); three (8%) patients experiencing serious infections in setting of multiple comorbidities requiring systemic intervention and hospitalization. Our low infection rate aligns with the known persistence of protective TRM cells in the skin and presumably through non-lymphoid and lymphoid tissues while on LDA therapy, thereby not interfering with immunosurveillance [11, 31, 32]. While higher infection rates are seen in patients treated with high-dose alemtuzumab, studies have not found significant differences in the effectiveness of higher dose regimens for CTCL [15]. While evaluating the safety of LDA in CTCL has been limited by relatively small numbers of patients treated, significant safety data exists using higher doses in treating multiple sclerosis where rates of serious infections are 2–4% [33].

Intense and unremitting pruritus is experienced in an estimated 62–77% of patients with CTCL and is an enormous source of emotional and psychological distress [34]. In our study, LDA alleviated itching in a decisive manner; pruritus abated as soon as 1 month in most patients, with cessation of itching in all patients with no history of PPT at approximately 3 months from first LDA treatment. LDA reduced itching in patients within both clinical phenotypes (those with TCM and TRM/MM) in a short amount of time, although to a lesser extent in those with the TRM clinical phenotype. Patients with TCM phenotype were significantly more likely to cease itching than patients with TRM/TMM phenotype. Similarly, in a phase II study of alemtuzumab in advanced MF/SS, itching decreased from a median of 8 before treatment to 2 after treatment (based on a self-assessment of overall itch, on a 0 “no itch” to 10 “worst itch imaginable” scale) [35]. This reveals a dramatic association between circulating malignant T-cells and clinical pruritus.

In summary, our study supports the use of LDA in the treatment of CTCL with blood involvement, particularly in patients with the appropriate clinical phenotype – those presenting with blanching erythema and no active or prior palpable disease – the TCM phenotype. To our knowledge, this is the largest cohort of CTCL patients treated with LDA with a minimum of 5 years of clinical follow-up. Our patients with a history of PPT also responded to LDA but to a lesser extent; of these patients, those with no palpable disease at the time of LDA treatment had better response than with PPT. We also found a low rate of serious infections among our cohort of patients and that the 5-year incidence of death in the LDA treated cohort significantly improved compared to a crude measurement of incidence of death from reported OS for patients with advanced-stage CTCL prior to the use of alemtuzumab [36]. Finally, CR or PR induced by alemtuzumab was associated with a dramatic decline in pruritus, one of the most refractory symptoms of this disease [37]. Our study is limited by its retrospective nature, is limited to a single tertiary academic referral center and by its small sample size; however, the small sample size reflects the low prevalence of these NHLs and is in keeping with other studies of TCLs. To date, this is the largest retrospective cohort evaluating LDA for CTCL in the U.S.A. With the ongoing need for a tolerable and effective therapy for advanced-stage CTCL, LDA produces promising clinical responses in appropriate patients.