Usefulness of KIR3DL2 to Diagnose, Follow-Up, and Manage the Treatment of Patients with Sézary Syndrome

Sézary syndrome is a rare, aggressive, and leukemic form of cutaneous T-cell lymphoma (CTCL) characterized by erythroderma associated with generalized peripheral lymphadenopathy and circulating clonal malignant T cells called Sézary cells (1). Although the incidence of CTCL is 7.7 new cases per million persons per year, Sézary syndrome accounts for only 5% of these cases (2). Patients with Sézary syndrome have a bad prognosis, with a 5-year overall survival varying from 24% to 43% (3, 4). Current prognostic factors of Sézary syndrome include advanced age, elevated level of lactate dehydrogenase (LDH), raised count of eosinophils, a former history of mycosis fungoides, extent of peripheral blood involvement, and partial or total loss of CD26 and/or CD7 expression by circulating CD4⁺ T cells (5–12). There is no curative treatment and available systemic treatments have often short-lived responses, with relapses after few weeks or months (13). In the current International Society for Cutaneous Lymphomas (ISCL)/European Organisation for Research and Treatment of Cancer (EORTC) TNMB staging classification, the diagnosis of Sézary syndrome requires erythroderma with a positive T-cell clone in the peripheral blood associated with at least one B2 criterion including the identification of more than 1,000 Sézary cells/mm³ in the blood as determined been cytomorphologic analysis (14). Indeed Sézary cells have first been described as large atypical mononuclear cells with cerebriform nuclei (15). However, the detection of Sézary cells by cytomorphology is not absolutely specific for the diagnosis of Sézary syndrome, as such cells can be found in other inflammatory erythrodermic conditions (16).

KIR3DL2 is a member of the family of the killer-cell immunoglobulin-like receptors that was initially identified as expressed on natural killer (NK) cells and on rare circulating CD8⁺ T cells (17). KIR3DL2 has first been identified on the surface of Sézary cells in 2001, its sensitivity to establish the diagnosis of Sézary syndrome being confirmed thereafter in several studies (18–21). Indeed, it has been shown that the circulating CD4⁺KIR3DL2⁺ T cells corresponded to the clonal malignant T-cell population identified by its unique TCR-Vβ rearrangement (22, 23).

The aim of our study was to evaluate the expression of KIR3DL2 on peripheral blood T cells as a diagnostic, prognostic, and follow-up marker of Sézary syndrome as compared with the classical clinical and biological criteria, on a well-defined and large cohort of patients with Sézary syndrome with an 8-year follow-up.

A total of 64 patients diagnosed with Sézary syndrome and followed up in our referral center for cutaneous lymphoma (Saint-Louis Hospital, Paris, France) between October 2007 and April 2016 were included in the study. The patients were diagnosed according to the ISCL/EORTC classification for primary CTCL. Because a lymph node biopsy was not systematically performed at diagnosis for all patients, disease staging was available only for half of them and was as follows: stage IVA1 (n = 16), stage IVA2 (n = 13), and stage IVB (n = 3). Further detailed patients' characteristics are shown in Table 1. The following clinical features were assessed: age, sex, former diagnosis of CTCL, and treatments. The data were collected at diagnosis, after each change of treatment line and in some cases during the course of the treatment. Date of death or clinical status at the end of the study was collected. Clinical response to each treatment was evaluated according to the modified Severity Weighted Assessment Tool (mSWAT) score (24). Complete response (CR) was defined as 100% clearance of skin lesions, partial response (PR) as 50% to 99% clearance of skin disease from baseline without new tumors, stable disease (SD) as <25% increase to <50% clearance in skin disease from baseline without new tumors, progressive disease (PD) as ≥25% increase in skin disease from baseline or new tumors increase of skin score of greater than the sum of nadir plus 50% baseline score in patients with CR or PR, and relapse as any disease recurrence in those with CR.

Circulating Sézary cell (CSC) count, CD4/CD8 ratio, percentages of CD4⁺CD7⁻ and CD4⁺CD26⁻ T cells among CD4⁺ T cells, eosinophil cell count, and LDH levels were evaluated at diagnosis and during follow-up. The CSC count was assessed by expert investigators, on a morphologic basis, and expressed as absolute Sézary cell count/mm³. A raised eosinophil cell count was defined as greater than 700/mm³ and raised LDH value as greater than 450 U/L.

The same day, the percentages of lymphocytes (including T, B, and NK cells), CD3⁺CD4⁺ T cells among lymphocytes, and KIR3DL2-positive cells within the CD3⁺ and CD4⁺ T-cell subsets respectively were determined, either on total blood or/and peripheral blood mononuclear cells (PBMC) isolated by density gradient centrifugation. Four-color immunofluorescence staining was performed using fluorescein isothiocyanate–conjugated (FITC), R-phycoerythrin (PE), Phycoerythrin-Cyanin 5 (PC5), and Phycoerythrin-Cyanin 7 (PC7) anti-CD3, anti-CD4, anti-CD56, and anti-KIR3DL2 mAbs. KIR3DL2 mAbs used were as follows: Q66 (2010–February 2012; kindly gifted by Dr A. Moretta, Genova, Italy), 15C11 (March 2012–June 2012), and 13E4 (July 2012–April 2016; both produced and provided by Innate Pharma, Marseille, France). Cells were analyzed on a FC500 cytometer (Beckman Coulter) and the recorded data analyzed with FlowJo software. All measurements were done after obtaining patient written informed consent.

On the basis of previous studies, positivity was determined by more than 5% of CD3⁺CD4⁺KIR3DL2⁺ T lymphocytes among the CD4⁺ T-cell population (17, 22). Because some Sézary cells may be CD8⁺ T cells or have lost expression of CD4, positivity of KIR3DL2 was also monitored on the CD3⁺ CD56⁻ T-cell population when necessary (25, 26).

All data are presented as median (range) for continuous variables and numbers, and as percentages for categorical variables, unless stated otherwise. The χ² value was used to compare categorical variables. Receiver-operating characteristic (ROC) curves were designed to define the best value of CD4⁺KIR3DL2⁺ T-cell percentage among total CD3⁺CD4⁺ T lymphocytes to predict the risk of death at 2 years. The correlation between KIR3DL2 expression, CD26 loss, and TCRVβ-positivity on CD4⁺ T cells, all evaluated by flow cytometry, was analyzed with a nonparametric Spearman correlation test. For survival analysis, disease-specific survival was calculated from date of diagnosis to date of death as a result of Sézary syndrome or treatment secondary effects until April 2016. The Kaplan–Meier method with log-rank test was used to compare survival curves in two different groups, and multivariable Cox regression was used to analyze the independent effect of several risk factors on survival. P values less than 0.05 were considered significant. Analyses were performed with MedCalc version 16.2.1 software.

Between October 2007 and April 2016, for all patients diagnosed with Sézary syndrome and at each visit, a blood sample was processed to evaluate the percentage of KIR3DL2⁺ cells within total CD3⁺ or CD3⁺CD4⁺ T-lymphocyte population by flow cytometry. Following data collection, we first studied the prevalence of KIR3DL2-positive cells among the circulating CD4⁺ T-cell population at diagnosis and compared with the classical diagnostic markers of Sézary syndrome (Table 1). Note that the positivity threshold for KIR3DL2 was previously fixed at 5%, corresponding to the mean percentage of CD4⁺KIR3DL2⁺ T cells found in healthy donors or patients with benign erythroderma cohorts (17, 22). We found that 54 patients out of 64 tested (84.4%) had abnormal percentages (≥5%) of KIR3DL2-positive CD4⁺ T cells (range, 7.0%–98.1%). In addition, two patients showed KIR3DL2 expression on an abnormally elevated population of CD3⁺CD4⁻ T cells, these cells being lately identified as the malignant T-cell clone by anti-TCRVβ immunostaining (data not shown). For the eight remaining patients, KIR3DL2 expression was undetectable while other diagnostic factors were met [erythroderma and positive T-cell clone in the peripheral blood associated with elevated CSC count (>1,000/mm³) and CD4/CD8 ratio (>10)]. No significant correlation was found between the percentage of circulating CD4⁺ KIR3DL2⁺ cells and the disease stage at diagnosis, when determined, or the age of the patient (Table 2). In total, 87.5% of the patients (56/64) showed KIR3DL2 positivity at diagnosis. In comparison, evaluation of the tumor cell burden by CD26 staining showed a significant loss of CD26 marker on the CD4⁺ T-cell population for 18 of the 22 patients (81.8%) for whom this criterion was evaluated at diagnosis (Table 1; Fig. 1A). Note that CD26 negativity and KIR3DL2 positivity on CD4⁺ T cells were similarly correlated to the percentages of tumor cells identified by TCRVβ staining in the patients for whom all three parameters were available (Spearman's coefficient correlation = 0.37 and 0.38, respectively; n = 10; Fig. 1B and C). Nevertheless, our results make of KIR3DL2 expression a very sensitive diagnostic factor for Sézary syndrome as compared with the lack of CD26 expression.

In the entire group of patients, an univariate analysis established that an increased eosinophil cell count over 700/mm³, an elevated LDH level of more than 450 U/L, and a proportion of KIR3DL2⁺ cells within the CD3⁺ T cells superior to 85% at diagnosis were associated with a significant reduced disease-specific survival (DSS; Table 2; Fig. 2). A multivariate analysis showed that a percentage of KIR3DL2⁺ CD3⁺ T cells over 85% and a raised eosinophil cell count (>700/mm³) were independent parameters for reduced DSS. There was a trend for considering a CSC count, when over 6,500/mm³, as predictive of reduced DSS but for our patients' cohort, this did not prove to be significant. Altogether these comparative results show that the evaluation of the tumor cell burden among T cells through the detection of KIR3DL2 expression is one of the most relevant prognostic factors for Sézary syndrome. Of note, among the eight Sézary patients having normal amounts of circulating KIR3DL2⁺ CD4⁺ T cells (<5%) at diagnosis, with a 2.5 years of median follow-up, only one died 9 months after diagnosis. Although this low number of patients showing undetectable KIR3DL2 expression on their malignant clone does not allow statistical analysis, it demonstrates a better survival tendency for patients with such cellular phenotypic characteristic than for patients whose tumor cells display high levels of KIR3DL2.

To further evaluate the accuracy of KIR3DL2 as a follow-up marker of the circulating tumor cell burden, a comparison was made between the detection of KIR3DL2⁺CD4⁺ T cells and CSC counts over time. Interestingly, although CSC became undetectable in 12 patients during the follow-up, significant amounts of KIR3DL2⁺ CD4⁺ T cells always remained detectable at the same time, with percentages ranging from 10.7% to 98.1%, except for patients experiencing a severe lymphopenia upon alemtuzumab (anti-CD52 mAb) treatment (n = 2). This observation most likely reflected either a cell shape change (transition from enlarged to normal-sized lymphocytes) upon treatment with no reduction of the circulating tumor burden (n = 6; range of KIR3DL2⁺ CD4⁺ T cells: 10.7–98.1%; see data of a representative patient in Fig. 3A), or a decrease in the amount of malignant T cells but with the persistence of a residual population of CD3⁺CD4⁺KIR3DL2⁺ tumor T lymphocytes (n = 4; range of KIR3DL2⁺ CD4⁺ T cells: 12–72.8%; exemplified in Fig. 3B). Both situations resulted in a null CSC count, the tumor cells not fulfilling the cytomorphologic criteria anymore. Remarkably, clinical and biological relapse of the Sézary syndrome finally occurred in these patients, suggesting that KIR3DL2 might represent a powerful predictive relapse marker.

In our cohort, a decrease greater than 50% of CSC estimated by either cytomorphology or absolute KIR3DL2⁺ CD4⁺ T cells was statistically associated with PR or CR of the Sézary syndrome (P = 0.006 and 0.001, respectively). In contrast, PR or CR was not statistically associated with a stabilization of the percentage of KIR3DL2-positive cells among the CD4⁺ T-cell subset (P = 0.26). This is illustrated by the results obtained during a 5-year follow-up of a bexarotene-treated patient (Fig. 4A). An initial benefit of the treatment was detected during the first year, with a decrease in both CSC counts and circulating KIR3DL2⁺ CD4⁺ T cells. This period of improvement was then followed by a phase where CSC counts, total lymphocytes, and CD3⁺CD4⁺ T cell levels remained rather stable, whereas KIR3DL2-positive T cells reappeared, suggesting a loss of treatment efficacy with a contained but still active malignant cell resurgence despite PR. Remarkably, the malignant T cells gradually became the dominant cell type within total lymphocytes, with no detection of increased lymphocyte count, increased CSC count, or clinical signs of relapse. This increase in the percentage of KIR3DL2⁺ cells within the CD4⁺ T-cell population largely preceded the detection and increase of CSC count and the occurrence of PD, the latter being declared 1 year later when all clinical parameters converged toward disease progression. Such discrepancy was observed for nine patients who presented a percentage of KIR3DL2⁺ CD4⁺ T cells of 10.1% to 93.2% (median: 75.5%), whereas CR was declared according to clinical parameters.

In addition, the evaluation of KIR3DL2⁺CD4⁺ T-cell content could also enable the monitoring of the real impact of the treatment on the malignant T-cell clone. Thus, although a decrease in total lymphocytes count might suggest an improvement regarding blood tumor burden, it might not be immediately correlated with a reduction of the amount of malignant CD4⁺ T cells, suggesting that lymphocyte depletion occurred regardless of their malignant status. Although an average 50% reduction in the lymphocyte blood content was detected for 41 patients following treatment, for 10 of them this was not correlated with a significant decrease in their % of circulating KIR3DL2⁺ CD4⁺ T cells (range, 45.6%–98.1%; see Fig. 4B for a representative example). Eventually, tumor cell depletion was partially achieved, with a return to normal lymphocytes levels, but again malignant cells persisted there and relapse finally occurred. KIR3DL2 detection therefore permitted to estimate whether the ongoing treatment specifically targeted the malignant T-cell clone and if so, to visualize the pool of residual tumor cells. Altogether, these data support that KIR3DL2 detection might therefore be safely used to assess treatment efficiency and more importantly represent a very early marker predictive for relapse and progression.

Among our cohort of patients with Sézary syndrome, six were treated with the anti-CD52 antibody alemtuzumab. As exemplified with a representative patient (Fig. 4C), alemtuzumab administration led to an overall lymphopenia, a target-specific side effect of this immunotherapy. However, upon renewal of the lymphocytes population, malignant KIR3DL2⁺ CD4⁺ T cells became readily detectable and represented most of the CD4⁺ T cells, although no increased lymphocyte count was engaged yet. This represents the first demonstration that malignant T cells immediately reappear upon reconstitution of the circulating lymphocytes population despite prior total elimination of the T-cell subset.

This study, performed on 64 patients with Sézary syndrome diagnosed and followed-up in our referral center for cutaneous lymphoma (Saint Louis Hospital, Paris, France), is the first to demonstrate that KIR3DL2 is not only the most sensitive marker for Sézary syndrome diagnosis as compared with all other available paraclinical criteria of Sézary syndrome, but is also the best independent prognostic factors for death in this disease. In addition, it appears to be an accurate and very early marker to detect the residual disease and the risk of disease relapse.

KIR3DL2 is a recently discovered marker of circulating and cutaneous malignant T lymphocytes in Sézary syndrome (18). It has been previously showed as useful to evaluate the tumor burden in peripheral blood both at diagnosis and during follow-up, as KIR3DL2⁺ T-lymphocyte count strongly correlates with the evaluation of CSCs obtained either by performing a TCR-Vβ determination by flow cytometry or by a cytomorphologic analysis (20–23). We here demonstrate that KIR3DL2 is the most sensitive diagnostic criteria for Sézary syndrome. Although for some patients KIR3DL2 may not be detected on the malignant T-cell clone, most likely reflecting low levels of KIR3DL2 expression that were below the detection threshold of our standard flow cytometry method and/or insufficient anti-KIR3DL2 antibody affinity, the immunostaining of KIR3DL2⁺ T cells allows a better evaluation of the circulating tumor burden than the traditional Sézary cell count established by cytomorphology, and may help to establish the malignant character of peripheral blood lymphocytes in otherwise unclear cases. Indeed, although Sézary-like cells can be observed by cytomorphologic analysis in other inflammatory skin conditions, KIR3DL2 expression appears highly restricted to Sézary tumor clone (22, 23, 27). In addition, in our daily practice, we have noticed that the detection of KIR3DL2 positivity among CD4⁺ T cells allowed an early diagnosis of Sézary syndrome for erythrodermic patients who do not present the classically admitted diagnostic markers (28). Indeed, in these patients, positivity for KIR3DL2 may occur before the appearance of the other classic diagnostic markers. Moreover, the assessment of the percentage of KIR3DL2⁺ cells among the CD4⁺ T cells by flow cytometry allows a fast, easy, and accurate determination of the tumor lymphocyte rate among total lymphocytes in peripheral blood. Importantly, KIR3DL2 immunostaining enables the detection of a malignant clone that may not be identified through the use of the commercially available anti-TCR-Vβ mAb panel (that only covers about 70% of normal human TCR-Vβ repertoire; ref. 22) or that may show a complete downmodulation of pan-T antigens such as CD4. Finally, although until now the detection of Sézary cells by flow cytometry essentially relies on the loss of surface antigen expression (e.g., CD26, CD7, CD4), KIR3DL2 represents the first positive marker allowing an identification of the tumor cells.

Few studies have specifically evaluated the prognostic factors for Sézary syndrome only, as it is a rare condition, studies often gathering Sézary syndrome and mycosis fungoides patients together. The main adverse prognostic factor in Sézary syndrome has been shown to be the extent of peripheral blood involvement evaluated by CSC count based on morphologic criteria, but this parameter was not statistically significant in our study (9, 12). In contrast, we found that an increased eosinophil count and an increased LDH level at diagnosis were important prognostic factors in univariable analysis for patients with Sézary syndrome. Age has also been found as a major independent prognostic factor in many studies but did not reach statistical significance for our patients' cohort (6, 10, 12). This might be explained by the use of more aggressive treatments, associated to a high risk of treatment-related mortality in younger patients. In a multivariate analysis, among the classic prognostic factors evaluated in Sézary syndrome, only blood eosinophilia at diagnosis was an independent prognostic factor for disease-specific death. In contrast to a previous study, we did not find that a previous diagnosis of mycosis fungoides was an unfavorable variable (5). Another prognostic factor was the loss of CD26 or/and CD7 expression, unfortunately we could not analyze this prognostic value, as it was not evaluated for all our patients at diagnosis (7, 8). However, we clearly establish that KIR3DL2 is the main independent prognostic marker for disease-specific death, as patients having more than 85% of KIR3DL2-positive cells among their circulating T cells at diagnosis had a significant decrease in their life expectancy.

We further show that the absolute CSC, determined by cytomorphology, is of less relevant value for prognosis than the percentage of CD4⁺ KIR3DL2⁺ T cells among peripheral blood lymphocytes. Indeed, even when patients receive efficient treatment against Sézary syndrome and experience complete or partial remission of the disease, with decreased or negative absolute CSC count evaluated by cytomorphology, the proportion of malignant lymphocytes evaluated by KIR3DL2 positivity may show relevant residual or progressive disease. This most likely reflects the fact that malignant T cells do not always present an abnormal morphology while constantly expressing KIR3DL2. In addition, the detection of the KIR3DL2⁺ malignant cells during treatment provides valuable information even when these cells are reduced to a minor population or/and in absence of an increased lymphocyte count. It further enables to distinguish treatment-induced effects on nonmalignant lymphocytes from effects on the malignant cell burden. In this regard, we experienced that an unspecific depletion of normal lymphocytes often precedes the visible tumor T-cell burden reduction. Conversely, we demonstrated that the reamplification of the KIR3DL2⁺ CD4⁺ T-cell population within T cells could occur whereas disease progression, relapse, or flare was not assessed by classically admitted parameters. This latter observation clearly promotes the use of KIR3DL2 as a very early tumor marker and for the surveillance of the circulating tumor burden during and after treatment even though CR or PR, evaluated by classical disease parameters, seems to be achieved. It might indeed represent a valuable tool for disease follow-up especially when considering that most treatments do not result in a complete and definitive depletion of the tumor cells and that the persistence of residual malignant cells among apparently normal lymphocytes is a warning sign for relapse or lymphoproliferative flare.

Finally, the follow-up data from alemtuzumab-treated patients demonstrated that a complete depletion of lymphocytes is not associated with durable remission. Beside the fact that such treatment induced severe immunosuppression, which is at high risk of community-acquired, nosocomial, and opportunistic infections (all of them being identified as the main cause of death in patients with Sézary syndrome; refs. 29–31), it does not prevent the reappearance of tumor cells that occurred concomitantly to the lymphocytes subset reconstitution. Although the precise origin of the regenerating tumor burden has to be identified, KIR3DL2 marker remains detectable and could be helpful to discriminate between short- or long-term efficiency of the treatment. More widely, our data favor the use of KIR3DL2 determination as a reliable indicator of therapeutic effectiveness that could greatly shortened the time lapse needed to ascertain resistance or relapse, and consequently treatment adjustment.

In conclusion, we show that KIR3DL2 represents a valuable tool for a routine use as a clinical parameter at diagnosis, for prognosis, and to assess the treatment efficiency for Sézary syndrome. Besides its use as a specific marker for CTCL malignant clone, KIR3DL2 might also represent an attractive therapeutic target for this disease. In this regard, promising preclinical data were obtained by using an anti-KIR3DL2 humanized antibody able to promote both antibody-dependent cell cytotoxicity and phagocytosis, and consequently a specific depletion of the tumor cells (22, 32). Such approach sounds promising as minimal side effects would be expected when considering the antibody modes of action.

M. Bagot holds ownership interest (including patents) in Innate Pharma, and is a consultant/advisory board member for Takeda. No potential conflicts of interest were disclosed by the other authors.

Conception and design: C. Hurabielle, A. Bensussan, M. Bagot, A. Marie-Cardine

Development of methodology: N. Thonnart, M. Bagot, A. Marie-Cardine

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C. Hurabielle, N. Thonnart, C. Ram-Wolff, M. Bagot, A. Marie-Cardine

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C. Hurabielle, M. Bagot, A. Marie-Cardine

Writing, review, and/or revision of the manuscript: C. Hurabielle, C. Ram-Wolff, H. Sicard, A. Bensussan, M. Bagot, A. Marie-Cardine

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): H. Sicard, M. Bagot

Study supervision: A. Bensussan, M. Bagot, A. Marie-Cardine

This work was supported by funding from the French National Institute for Health and Medical Research (INSERM), INCa-DGOS, and Paris Diderot University.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.