Using next-generation sequencing (NGS), we recently documented T-cell oligoclonality in treatment-naïve chronic lymphocytic leukemia (CLL), with evidence indicating T-cell selection by restricted antigens.
Here, we sought to comprehensively assess T-cell repertoire changes during treatment in relation to (i) treatment type [fludarabine-cyclophosphamide-rituximab (FCR) versus ibrutinib (IB) versus rituximab-idelalisib (R-ID)], and (ii) clinical response, by combining NGS immunoprofiling, flow cytometry, and functional bioassays.
T-cell clonality significantly increased at (i) 3 months in the FCR and R-ID treatment groups, and (ii) over deepening clinical response in the R-ID group, with a similar trend detected in the IB group. Notably, in constrast to FCR that induced T-cell repertoire reconstitution, B-cell receptor signaling inhibitors (BcRi) preserved pretreatment clones. Extensive comparisons both within CLL as well as against T-cell receptor sequence databases showed little similarity with other entities, but instead revealed major clonotypes shared exclusively by patients with CLL, alluding to selection by conserved CLL-associated antigens. We then evaluated the functional effect of treatments on T cells and found that (i) R-ID upregulated the expression of activation markers in effector memory T cells, and (ii) both BcRi improved antitumor T-cell immune synapse formation, in marked contrast to FCR.
Taken together, our NGS immunoprofiling data suggest that BcRi retain T-cell clones that may have developed against CLL-associated antigens. Phenotypic and immune synapse bioassays support a concurrent restoration of functionality, mostly evident for R-ID, arguably contributing to clinical response.
Chronic lymphocytic leukemia (CLL) T cells display largely defective antitumor responses. From a clinical perspective, there is a clear need to optimize CLL immunotherapy, especially regarding minimal residual disease eradication. Boosting antitumor T-cell responses could ideally fit this gap, and identifying the best combination treatment partners is of translational relevance. We investigated the effect of treatment on CLL T cells in relation to (i) treatment type (FCR vs. ibrutinib vs. rituximab-idelalisib), and (ii) clinical response. In contrast to chemoimmunotherapy, B-cell receptor signaling inhibitors (BcRi) preserved pretreatment T-cell clones, which expanded further as treatment continued and the clinical response deepened. We found major clonotypes shared among different patients, raising the possibility of selection by conserved, CLL-associated epitopes. A concurrent restoration of T-cell functionality with BcRi therapy arguably contributed to clinical response. Overall, these data provide a rationale for designing combination strategies aiming to boost cytotoxic antitumor responses.
The tumor microenvironment (TME) plays a pivotal role in the natural course of chronic lymphocytic leukemia (CLL). This is evidenced by ample immunogenetic cues suggesting selection of the malignant B-cell clone by a restricted set of antigens, as well as by the remarkable clinical efficacy of drugs interfering with B-cell receptor signaling (BcR signaling inhibitors, BcRi; refs. 1–6). Although BcRi were developed to target the clonal B cells, accumulating data suggest a pleiotropic effect in the CLL microenvironment, particularly T cells, which may contribute to their clinical efficacy (7–15).
T cells are intimately implicated in CLL pathophysiology. Xenograft studies have highlighted the importance of trophic signals provided by T cells for the survival of the malignant clone (16). On the other hand, immunosurveillance by T cells is compromised in CLL, due to, among others, defective cytoskeletal signaling and lytic immune synapse formation following tumor immunosuppressive signaling (17–20). Therefore, delineating the mechanisms driving T-cell tolerance rather than effective antitumor cytotoxic immune responses is highly relevant and could facilitate optimization of immunotherapy for CLL, with the aim of eradicating minimal residual disease (21, 22).
Studies employing different methodologies have demonstrated the existence of T-cell expansions in CLL (23–27). Recently, we investigated the role of antigenic stimulation in shaping the T-cell repertoire in treatment-naive CLL using next-generation sequencing (NGS; ref. 28). Our study confirmed T-cell oligoclonality, with major (high-frequency) clonotypes persisting and further expanding overtime as well as clonotypes shared among different patients, which appeared to be CLL-specific. Overall, these findings pointed toward selection of T cells by CLL-associated antigenic epitopes.
On these grounds, we sought to comprehensively assess the effect of treatment on CLL T-cell repertoire dynamics by combining NGS immunogenetics, flow cytometry and immune synapse bioassays, to investigate associations of quantitative and/or qualitative changes with the type of treatment and clinical response. Considering recent data about collateral effects of BcRi treatment on T cells (7–14, 29), we focused on patients receiving ibrutinib or rituximab-idelalisib which we compared to patients receiving standard chemoimmunotherapy with the fludarabine-cyclophosphamide-rituximab regimen.
In sharp contrast to chemoimmunotherapy, BcRi were found to preserve pretreatment T-cell clones. These clones expanded further as treatment continued and clinical response deepened, particularly in the case of rituximab–idelalisib. Repertoire comparisons revealed major clonotypes shared by different patients with CLL but very few “public” clonotypes (i.e., also present in other conditions), suggesting selection by conserved CLL-associated epitopes. Moreover, phenotypic and immune synapse bioassays supported a concurrent restoration of T-cell functionality with BcRi therapy, mostly evident for rituximab–idelalisib, arguably contributing to clinical response. Taken together, these data provide a rationale for designing combination strategies aiming to boost cytotoxic antitumor responses.
Materials and Methods
We analyzed samples from 28 patients with CLL treated in two centres (Thessaloniki, Greece; Milan, Italy) with FCR (n = 9), IB (n = 15), and/or rituximab–idelalisib (n = 10). Patients were selected on the basis of sample availability. Demographic and clinicobiological characteristics are provided in Supplementary Table S1. For 22 patients, samples were collected over the first line of treatment, either fludarabine–cyclophosphamide–rituximab or BcRi. For the remaining 6 of 28 patients, we analyzed samples collected over consecutive lines of treatment (first-line fludarabine–cyclophosphamide–rituximab/second-line ibrutinib, n = 3; first-line FCR/second-line rituximab–idelalisib, n = 2; first-line IB/second-line rituximab–idelalisib, n = 1). In all cases, samples posttreatment initiation were analyzed in relation to a respective pretreatment initiation sample. In total, we analyzed 98 peripheral blood (PB) samples: (i) for patients who received fludarabine–cyclophosphamide–rituximab, we analyzed samples collected pretreatment (n = 9) and at 3 months after the completion of the 6 cycles (n = 9, all in complete remission); (ii) for patients who received BcRi, we analyzed samples pretreatment (n = 25), at 3 months (n = 25), 9 months (n = 17), best clinical response (median 20 months, n = 13)] while still on continuous BcRi treatment. Best response timepoint was defined by the attending physician, and corresponded to the maximal duration of BcRi treatment at the time the study was conducted (median time on BcRi treatment: 20 months), with the single exception of a patient who progressed on ibrutinib and proceeded to rituximab-idelalisib as second-line treatment. This patient was assessed only at the 3-month timepoint while on ibrutinib, at which he had achieved PR. Also, we included 6 bone marrow (BM) samples, collected along with the respective PB pretreatment sample. A schematic representation of study cohort is provided in Fig. 1A.
No patient had evidence of infection at sampling. All patients were negative for HBV DNA, anti-HCV, and anti-HIV I/II. We carefully excluded patients who developed rituximab-related late-onset neutropenia (R-LON), as it is known to be mediated by cytotoxic T-cell clones (30). The local Ethics Review Committee approved the study and written informed consent form was obtained from all individuals in accordance with the Declaration of Helsinki.
Next-generation sequencing, definitions, and interpretation
TRBV–TRBD–TRBJ gene rearrangements were RT-PCR amplified and subjected to paired-end NGS (MiSeq, Illumina), as described previously (28). For all samples, the starting absolute T-cell count exceeded 0.5 × 106 cells to ensure adequate repertoire profiling depth.
Paired-end protocol allowed sequencing of the TRB complementarity-determining region 3 (CDR3) twice/read, thus increasing the accuracy of results. To further increase the accuracy of results, raw reads were processed through a purpose-built bioinformatics pipeline performing: (i) length/quality filtering of raw reads; (ii) merging of filtered-in paired reads via local alignment; (iii) length/quality filtering of stitched sequences (28). Detailed length, quality, and overlap rules are provided in Supplementary Table S2. Importantly, no base calls of Q-score <30 were allowed in the 75 nucleotide stretch preceding the FGXG motif at the start of TRB FR4, thus further increasing the CDR3 sequencing reliability. Filtered-in sequences were submitted to IMGT/HighV-QUEST (http://www.imgt.org), and metadata was processed by a validated bioinformatics algorithm designed for clonotype computation and repertoire analysis (Fig. 1B; refs. 28, 31). Only productive, in-frame TRBV–TRBD–TRBJ gene rearrangements with functional TRBV genes were included in the analysis. TRBV–TRBD–TRBJ gene rearrangements carrying TRBV genes with <95% germline identity were also discarded as sequences with unacceptable error rate, given the lack of somatic hypermutation in T cells.
Clonotypes were computed as unique pairs of TRBV genes and CDR3 amino acid sequences within a sample. Clonotypes were considered expanded when they contained ≥2 sequences, otherwise they were considered as “singletons.” The 10 most expanded clonotypes within a sample are referred to as “major.” The relative frequency of each clonotype/sample was calculated as the number of rearrangements corresponding to the clonotype divided by the total number of productive, filtered-in rearrangements for that particular sample.
For TRB gene repertoire analysis, clonotypes rather than single rearrangements were considered to avoid potential biases due to expansion following antigenic stimulation, that is, individual TRBV gene frequencies within a sample were calculated as the number of clonotypes using particular TRBV genes over the total number of clonotypes. Each category of samples (PBMCs, BM) was analyzed separately.
Interpatient and across entities clonotype comparison
We determined the major clonotypes of all CLL samples included in the study (n = 563 unique clonotypes, i.e., 10 major clonotypes of each sample removing duplicate values such as clonotypes which persisted in overtime analysis of the same patient or clonotypes shared among different patients) and compared them across CLL patients of this cohort, and against: (i) all T-cell clonotypes from our previous NGS study involving treatment-naive CLL patients (n = 1,105,728; ref. 28), (ii) T-cell clonotypes from 15 healthy donors (NGS study by our group, n = 573,651; ref. 32), (iii) HHV-4 (Epstein–Barr virus, EBV), HHV-5 (cytomegalovirus, CMV), and BK virus–specific T-cell clonotypes (antiviral T-cell NGS study by our group, n = 169,502; ref. 33), (iv) nonredundant, well-annotated, unique T-cell clonotypes retrieved from public databases (n = 17,024; refs. 34, 35), and (v) T-cell clonotypes from previous low-throughput immunoprofiling studies (classic subcloning followed by Sanger sequencing) in clinical monoclonal B-cell lymphocytosis (high-count MBL, n = 545; ref. 36), as well as clinical entities associated with clonal T-cell expansions [R-LON (n = 283; ref. 30), chronic idiopathic neutropenia (CIN, n = 576; ref. 37), large granular T-cell leukemia (T-LGL, n = 932); refs. 38–40].
Multicolor flow cytometry (FACSCanto II cell analyzer, BD Biosciences) was performed to investigate the expression of activation and exhaustion markers (CD25, CD38, CD69, HLA-DR, and PD1, respectively) in functionally distinct T-cell subpopulations, namely: (i) CD4+ and CD8+ naive T cells (CD45RA+/CCR7+), (ii) CD4+ and CD8+ central memory T cells (CM, CD45RO+/CCR7+), (iii) CD4+ and CD8+ effector memory T cells (EM, CD45RO+/CCR7−), and (iv) CD4+ and CD8+ terminal effector memory RA cells (TEMRA, CD45RA+/CCR7−). Eight patients (rituximab–idelalisib, n = 4; ibrutinib, n = 4) pretreatment and at the 3-month timepoint were analyzed, as shown in Fig. 1A.
Viably frozen peripheral blood mononuclear cells were thawed and resuspended in PBS (15 × 106 cells in 1 mL) and incubated with pretitrated antibodies to cell-surface markers at 4°C for 20 minutes. CD45 V500, CD3 V450, CD4 PerCP, CD4 PE-Cy7, CD8 PE-Cy7, CCR7 APC-Cy7, CD45RO FITC, CD45RA FITC, CD69 PE, CD25 APC, CD38 PerCP, HLA-DR PE, and PD1 APC were used (BD Biosciences). Live lymphocytes were gated by forward and side scatter, CD45-positive plus 7AAD–negative staining. Data were analyzed using the Kaluza flow cytometry analysis software (Beckman Coulter; Supplementary Fig. S1).
An acquisition cut-off of 20,000 CD3+ live cells/sample was applied to ensure reliable representation of all studied T-cell subpopulations, as well as uniformity across samples. For T-cell activation/exhaustion analysis, results were expressed as the proportion of cells expressing antigens of interest (percent positive staining).
Immune synapse bioassays, confocal microscopy, and quantitative image analysis
Tumor cell–T-cell conjugate assays were designed to test the ability of CLL T cells to form immune synapses with autologous tumor cells. Negatively selected viable CD3+ T cells (1 × 106) from untreated, FCR-, or BCRi-treated patients were conjugated with equal numbers of autologous CLL tumor cells acting as antigen-presenting cells pulsed with super-antigen cocktail and stained with CMAC blue dye (ThermoFisher Scientific) according to the manufacturer's instructions. Cell conjugates were incubated for 15 minutes at 37°C, 5% CO2. Immunofluorescent labeling was done using Cytofuge2 cell concentrator, as described previously (19). Antibodies for rhodamine phalloidin (F-actin staining) and perforin (Alexafluor 488 conjugated) were applied for 45 minutes at 4°C in 5% goat serum (Sigma-Aldrich) solution. After washing, cell specimens were sealed with 22 × 32 mm coverslips using fluorescent mounting (Dako). Images were captured with an A1R confocal microscope (Nikon) using a 63 × oil objective with NIS- Elements software Version 5.01 imaging software (Nikon) and fluorescence was acquired sequentially. The specificity of staining was optimized using isotype-control antibodies. Blinded confocal images (n = 3 per treatment group) were analyzed using NIS-Elements software Version 5.01 to measure the F-actin (red fluorescent channel) and perforin (green fluorescent channel) polarization at contact sites and immune synapses using relative recruitment index (RRI) analysis. RRI is calculated as the ratio of F-actin/perforin mean fluorescent intensity at the immune synapse area to the respective value at a cellular membrane area not involved in an immune synapse (RRI<1 indicates a nonpolarized, dysfunctional immune synapse). RRI values were expressed as mean value ± SEM.
Statistical analysis and visualization tools
Descriptive statistics for discrete parameters included counts and frequency distributions. The significance of bivariate/multivariate relationships between variables was assessed using the Student t test, as well as the nonparametric Wilcoxon (for paired values) and Mann–Whitney tests. For all comparisons a significance level of P = 0.05 was set.
By applying the Wilcoxon-signed rank test for matched pairs, the power of the NGS analysis ranges from 92% (for the most populated comparison, i.e., pre-IB versus IB at 3 months, n = 15 samples) to 53% for the least populated comparisons, that is, pre- rituximab–idelalisib, rituximab–idelalisib at 3 months, rituximab–idelalisib at 9 months versus rituximab–idelalisib best response, n = 5 samples). A table with power values for each paired comparison is provided in Supplementary Material (Supplementary Table S3).
Data sharing statement
Raw sequence data have been submitted to the Sequence Read Archive (SRA); BioProject ID: PRJNA552792
The T-cell repertoire in CLL is skewed both pre- and posttreatment initiation in all treatment subgroups, with clonal expansions
First, we profiled by NGS the TR gene repertoire in terms of clonality and TRB gene usage. Overall, 23,262,732 TRBV-TRBD-TRBJ sequences were obtained, of which 20,347,768 (87.5%, median 155,479/sample) passed filters and were further analyzed. The median number of distinct clonotypes/sample was 11,420 (range: 2,672–54,045); the median number of expanded clonotypes and singletons/sample was 5,287 (range: 772–26,501) and 6,417 (range: 1500–32,428), respectively.
All CLL patient blood samples showed a T-cell oligoclonal profile. The median T-cell clonality of all analyzed samples, calculated as the cumulative frequency of the 10 major clonotypes/sample, was 37.5% (range 8.1%–70.5%).
Nine TRBV genes accounted for almost half of the total repertoire both pretreatment and at the 3-month timepoint (46.8% and 47.6% of the total repertoire, respectively). The relative frequency of each gene did not particularly differentiate over treatment, nor among treatment groups (fludarabine–cyclophosphamide–rituximab, ibrutinib, rituximab–idelalisib; Supplementary Fig. S2A). The single exception concerned the TRBV28 gene, which was underrepresented in the rituximab-idelalisib group both pretreatment and at the 3-month timepoint [3.0%/3.4% vs. 5.1%/5.1% in the FCR group (P < 0.01) vs. 4.5%/4.6% in the ibrutinib group (P < 0.01), respectively]. For both the ibrutinib and the rituximab–idelalisib groups, where the repertoire was studied beyond the 3-month timepoint, the TRBV gene frequency remained relatively stable overtime and with deepening clinical response (Supplementary Fig. S2B).
In conclusion, the T-cell repertoire was skewed in all examined subgroups, with restricted TRB gene usage and significant clonal expansions.
T-cell clonality increases over treatment and may be associated to clinical response
We then tested clonality dynamics over treatment and explored associations with clinical response. At cohort level (n = 28 patients), circulating T-cell clonality increased after treatment [median values from 30.3% pretreatment to 38.8% (P < 0.05), 41.9% (P < 0.05) and 46.2% (P < 0.01) at the 3-month, 9-month, and deepest clinical response timepoint, respectively].
Within different treatment groups, T-cell clonality increased at the 3-month timepoint in the case of fludarabine–cyclophosphamide–rituximab (from 28.8% to 46.9%, P < 0.05) and rituximab–idelalisib (from 33.0% to 39.1%, P < 0.01), but not ibrutinib (from 33.3% to 31.2%, P: 0.99). For the ibrutinib and rituximab–idelalisib groups, longitudinal analysis over treatment showed a gradual increase of T-cell clonality, related to the depth of clinical response, reaching statistical significance in the case of rituximab–idelalisib [from 33.0% pretreatment to 39.1% at 3 months (P < 0.01), 46.0% at 9 months (P < 0.01), and 46.1% at best clinical response (P: 0.13)], but not ibrutinib [from 33.3% pretreatment to 31.2% at 3 months (P: 0.99), 39.1% at 9 months (P: 0.44) and 42.1% at best clinical response (P: 0.17); Fig. 2].
We also performed analyses where clonality definition was modified in order to include a larger proportion of the repertoire (calculated as the cumulative frequency of the 20 or 50 most expanded clonotypes/sample, instead of 10). These analyses, as well as plots of clonotype frequency in relation to its rank in the repertoire, showed that the gradual increase of clonality resulted from the expansion of the major clonotypes, further alluding to their clinicobiological relevance (Supplementary Fig. S3).
Differential impact of chemoimmunotherapy versus BcRi on the T-cell repertoire
To further study clonal dynamics over treatment, we evaluated clonal persistence overtime for each treatment group. We found that chemoimmunotherapy (FCR) resulted in T-cell repertoire renewal through ablation and immune reconstitution, whereas BcRi retained most major pretreatment T-cell clones (median number of major pretreatment clonotypes that persisted at the 3-month timepoint: 2/10 for FCR versus 8/10 for ibrutinib vs. 7/10 for rituximab–idelalisib; Fig. 3A). Overlap analysis of the total repertoire pre- and posttreatment per group showed similar results (Fig. 3B). Furthermore, in the case of BcRi where the repertoire was studied longitudinally over treatment, repertoire preservation persisted over time (Fig. 3C).
Clonotype comparisons document the existence of “disease-biased” T-cell clonotypes
To obtain insight into the antigenic specificities of the expanded T-cell clones, we next listed all unique major clonotypes from our cohort (n = 563 clonotypes) and compared them (i) across patients, and (ii) against T-cell clonotypes from various entities, as detailed in the Materials and Methods section.
Thirty-two of the 563 clonotypes were shared among different patients with CLL (Supplementary Table S4), while 18 of 563 were found in another entity and/or healthy donors (“public” clonotypes”; Fig. 4). In particular, “public” clonotypes corresponded to 9 matches with T-cell clonotypes of herpes virus specificity (CMV, EBV), 2 matches with clonotypes of herpes virus specificity (EBV) that were also found in patients with CIN, and 5 matches with T-cell clonotypes from healthy donors. Importantly, of the 32 major clonotypes that were shared among patients with CLL, only 9 were “public,” whereas the remaining 23 (23/563, 4.1% of all major clonotypes of our cohort) were shared exclusively by patients with CLL, alluding to selection by CLL-associated antigens (“CLL-specific”). Of these 23 “CLL-specific” clonotypes, 3 were shared among 4 different patients, 5 among three different patients, and the remaining 15 among pairs of patients.
Major CLL-specific T-cell clonotypes expand further over treatment
Considering the above, we repeated our clonality analysis excluding all “public” clonotypes, with similar results. More specifically, clonality significantly increased at the 3-month timepoint in the case of FCR (from 20.8% to 46.9%, P < 0.01) and rituximab–idelalisib (from 32.9% to 38.8%, P < 0.01), but not ibrutinib (from 33.3% to 31.2%, P = 0.50). Overtime, clonality increased in both the rituximab–idelalisib and ibrutinib group, however, reaching statistical significance only in the former [rituximab–idelalisib: from 32.9% to 38.8% at 3 months (P < 0.01), 44.4% at 9 months (P < 0.01) and 39.5% at best clinical response (P: ns); IB: from 33.3% to 31.2% at 3 months (P:ns), 39.1% at 9 months (P: ns), and 42.1% at best clinical response (P: ns); Fig. 5].
Pretreatment major T-cell clonotypes from the peripheral blood are also found in the bone marrow
For 6 patients (rituximab–idelalisib, n = 3 and ibrutinib, n = 3), synchronous pretreatment PB and infiltrated BM samples were available. We performed pairwise comparison of the T-cell repertoires, to investigate for the presence of PB major clonotypes within the respective BM repertoire. Excluding “public” clonotypes, all PB major clonotypes of these patients (n = 56) were identified within the respective BM repertoires, and 37/56 (66.0%) were listed among the major clonotypes/BM sample.
Idelalisib-based therapy increases the expression of activation markers on effector memory T cells
NGS immunoprofiling showed that T-cell clonality increased with BcRis, with major clones persisting and further expanding over time. In some cases, these clones were shared among different patients with CLL but not found in other entities, suggesting that they may have developed in response to CLL-specific antigens. Therefore, we next wanted to investigate the impact of treatment on T-cell functionality, so as to uncover a possible link between these T-cell expansions and antitumor responses that could contribute to the deepening of clinical response.
To this end, we examined distinct T-cell subsets of 8 patients (rituximab–idelalisib, n = 4; ibrutinib, n = 4) pretreatment and at 3 months for the expression of activation and exhaustion markers by flow cytometry. While the relative frequencies of these T-cell subsets, namely CD4+ and CD8+ naïve, central memory (CM), effector memory (EM), and terminal effector memory (TEMRA) T cells, did not significantly fluctuate over treatment (data not shown), the expression of activation markers significantly increased posttreatment in the rituximab–idelalisib group compared with the ibrutinib group. More specifically, we noted increased expression of: (i) CD69 in CD4+ EM (median fold change 2.0 vs. 0.5, P < 0.01); (ii) CD25 in CD8+ TEMRA (median fold change 1.3 vs. 0.5, P < 0.01), and (iii) CD38 in CD8+ EM (median fold change 1.6 vs. 0.3, P < 0.05), and TEMRA T cells (median fold change 1.1 vs. 0.3, P < 0.05; Supplementary Fig. S4).
BcRi treatment improves T-cell immune synapse formation with tumor cells
Given that T-cell dysfunction in CLL has been linked to impaired immune synapse formation, we finally investigated the impact of therapy on immune function by characterizing the ability of patient T cells to form synapse interactions with autologous baseline CLL cells in 13 patients pretreatment and at 3 months (fludarabine–cyclophosphamide–rituximab, n = 3; rituximab–idelalisib, n = 5; IB, n = 5). Confocal microscopy with quantitative image analysis showed that T cells from FCR-treated patients did not improve their ability to form synapse interactions (Fig. 6A), whereas both BcRi increased the formation of polarized F-actin immune synapses with tumor cells (rituximab–idelalisib, ibrutinib, P < 0.05; fludarabine-cyclophosphamide-rituximab, P = 0.99; Fig. 6B and C). We also detected enhanced polarization of the cytolytic molecule perforin toward tumor cells in ibrutinib-associated T-cell synapses (P < 0.05; Fig. 6B). For 2 patients (rituximab–idelalisib, n = 1; ibrutinib, n = 1), RRI immune synapse analysis was also performed at the deepest clinical response timepoint. These assays revealed no change in T-cell synapse formation at the deepest response timepoint reached with ibrutinib therapy (25 months) compared with the 3-month timepoint, that may suggest a normalization of T-cell function. In contrast, for the patient who received rituximab–idelalisib, deeper clinical response (12 months) correlated with a further increase in polarized immune synapse function (Fig. 6D).
In this work, we sought to investigate the effect of different types of treatment on T cells from patients with CLL. Given the pleiotropic impact of BcRi on the CLL microenvironment, we focused on BcRi as currently used for CLL treatment (ibrutinib and rituximab–idelalisib) and compared with standard chemoimmunotherapy (fludarabine–cyclophosphamide–rituximab; ref. 13). Our study cohort was significantly enriched for CLL cases with unmutated IGHV genes, the main reason being that these cases are most frequently in need of treatment, especially BcRis. While it cannot directly account for IGHV-mutated CLL, all reported findings concern pair-matched samples (samples of the same patient over time), therefore we argue that IGHV mutational status is not an influent.
We report that T-cell clonality significantly increased after treatment with FCR and over treatment with rituximab–idelalisib, and, at least in the case of rituximab–idelalisib, this correlated with the depth of clinical response, peaking at 9 months. A similar trend was detected with ibrutinib therapy, however, an increase in clonality occurred later in the course of treatment (≥9 mo) and did not reach statistical significance, possibly due to the smaller number of samples at more advanced timepoints.
At first glance, the ibrutinib results may seem contradictory to a recent study by Yin and colleagues, reporting that ibrutinib therapy resulted in TR repertoire diversification; however, certain experimental and analytic differences may explain this (7). From the start, the use of different clonotype definitions renders the results incomparable: Yin and colleagues reported outcomes on the 1000 most frequent clonotypes (defined as productive unique sequences per sample), while in this study we focused on the 10 most frequent clonotypes, defined on the basis of unique amino acid CDR3 sequence and the same IGHV gene. Moreover, Yin and colleagues (7) utilized the ImmunoSEQ platform by Adaptive Biotechnologies, whereas we applied customized, purpose-built bioinformatics algorithms with intentionally strict length and quality filtering. Of note, the increasing T-cell repertoire diversity reported by Yin mostly results from an increasing number of singletons. When we performed the same type of analysis in our cohort, following clonotype definition as per the Yin and colleagues study, we also found an increase of the singleton fraction at the 9-month timepoint, although not reaching statistical significance. Besides, Yin and colleagues (7) mention that 6 of 16 dominant TRBV-TRBJ usages persist and further expand over 12 months of ibrutinib therapy, and therefore speculate antigenic stimulation, in an analogy to our results. Finally, Yin and colleagues (7) comment on repertoire diversity by focusing on the emergence of new low-frequency clones over treatment, while, we, on the other hand, focus on the dynamics of the most frequent clones on the consideration that these are the most biologically relevant, and dwell into the hypothesis that may have expanded in response to tumor antigens. We do not comment on repertoire diversity; this may be associated with infectious risks and complications, as hypothesized by Yin and colleagues, but is most probably irrelevant to the dynamics of antitumor clones.
Somewhat expectedly, fludarabine–cyclophosphamide–rituximab increased T-cell clonality through repertoire ablation and reconstitution, probably reflecting a lymphotoxic effect that is restored overtime. Indeed, in a patient who received FCR and was studied prospectively, the increase of clonality at 3 months posttreatment was restored to pretreatment levels by 9 months. In sharp contrast, BcRi did not reconstitute the T-cell repertoire, retaining pretreatment major T-cell clones during therapy, and this was shown consistently for the longest follow-up period reported so far (median 20 months).
To obtain evidence as to whether the herin identified T-cell clones might have expanded in response to CLL-associated antigens or not, we performed extensive comparisons against not only CLL/MBL sequence datasets (36, 41), but also datasets from common herpes virus infections (CMV/EBV), healthy individuals, and entities mediated by T-cell clones (30, 32–35, 37–40) and found a significant number of major T-cell clonotypes shared exclusively among patients with CLL. This is remarkable given: (i) the random HLA background of our cohort, and (ii) the size of the comparison dataset (761,968 distinct TRBV-TRBD-TRBJ sequences). Therefore, this finding supports the existence of (neo)epitopes that are conserved across CLL. Along the same line of reasoning, it is worth noting that, for 3 patients studied over sequential lines of treatment, we found major clonotypes prior to first-line FCR which disappeared at CR following FCR treatment, but reemerged as major clonotypes at first relapse and persisted through second-line treatment with BcRi (data not shown).
To investigate the functional impact of treatment on CLL T cells, we studied the expression of activation and exhaustion markers on T cells. We found that rituximab–idelalisib upregulates certain activation markers including IL2 receptor (CD25) in effector memory T cells compared with ibrutinib. Although this increase was arithmetically small, one needs to consider the fact that the frequency of each major clone is relatively low; therefore, even though we purposely focused our analysis on well-defined functional T-cell subpopulations, the immunophenotypic activation of antitumor T-cell clones may be “diluted” within the respective subpopulation. It is also possible that CLL T cells undergo phenotypic changes within tissue TMEs as well as the PB (42).
Considering that T-cell dysfunction in CLL manifests in a defective ability to form immune synapses with tumor B cells, we investigated the effect of treatment on immune synapse formation capability (18, 19). We found that, in contrast to fludarabine–cyclophosphamide–rituximab, BcRi improved immune synapse function and the recognition of tumor B cells. Interestingly, for the rituximab–idelalisib case which was studied beyond 3 months, this synapse restoration correlated with the depth of clinical response, however definite conclusions cannot be drawn as this was a single case. Improvement in immune synapse formation could be attributed to treatment-induced reduction in tumor load (as the T-cell defect has been shown to be tumor-induced; refs. 18, 19); however, the fact that this did not occur following fludarabine–cyclophosphamide–rituximab, the only treatment that led to complete remissions, argues in favor of a BcRi-specific effect on T-cell function. Admittedly, the number of cases analyzed for immune synapse formation was limited, due to the substantial number of T cells required for all assays performed in this study. Notably, our results are in keeping with a previous report studying the immunomodulatory impact of fludarabine–cyclophosphamide–rituximab and rituximab–ibrutinib (43).
Overall, NGS immunogenetics showed that BcRi retain T-cell clones which can expand in parallel to deepening clinical response and possibly contribute to it. This is further corroborated by phenotypic and functional bioassays that demonstrate concurrent T-cell activation and improvement in synapse recognition of tumor cells. Taken together, this study provides a rationale for designing future combinatorial therapeutic strategies aiming to boost antitumor immune responses. In support of this, recent studies have described the ability of both ibrutinib and idelalisib to enhance chimeric antigen receptor T-cell efficacy in CLL (44, 45). Future identification of the relevant CLL-associated antigens may eventually pave the way for stratified treatments by means of engineered T cells or peptide vaccines, especially if these epitopes are conserved among patients with CLL.
Disclosure of Potential Conflicts of Interest
A. Vardi reports grants from European Hematology Association (European Hematology Association Physician Scientist Research Grant) during the conduct of the study. P. Ghia reports grants and personal fees from AbbVie (advisory boards; research support), personal fees from Adaptive (advisory boards), personal fees from ArQule (advisory boards), personal fees from BeiGene (advisory boards), grants and personal fees from Gilead (advisory boards; research support), grants and personal fees from Janssen (advisory boards; research support), personal fees from Juno/Celgene (advisory boards), personal fees from Loxo (advisory boards), personal fees from MEI (advisory boards), grants and personal fees from Sunesis (advisory boards; research support), and personal fees from Acerta/AstraZeneca (advisory boards) outside the submitted work. K. Stamatopoulos reports grants from Janssen during the conduct of the study; grants and personal fees from AbbVie, personal fees from Janssen, and personal fees from Gilead outside the submitted work. A. Chatzidimitriou reports other from RANET Transcan-2, Project Acronym: Novel (supported in part) and other from Hellenic Precision Medicine Network in Oncology (supported in part) during the conduct of the study.
A. Vardi: Conceptualization, data curation, formal analysis, supervision, funding acquisition, investigation, visualization, methodology, writing-original draft, writing-review and editing. E. Vlachonikola: Data curation, formal analysis, investigation, methodology. D. Papazoglou: Investigation, methodology, writing-original draft. F. Psomopoulos: Software, visualization. K. Kotta: Investigation, methodology. N. Ioannou: Investigation, methodology, writing-original draft. C. Galigalidou: Investigation. K. Gemenetzi: Investigation. K. Pasentsis: Methodology. M. Kotouza: Software, visualization. E. Koravou: Investigation. L. Scarfó: Resources. M. Iskas: Resources. N. Stavroyianni: Resources. P. Ghia: Resources. A. Anagnostopoulos: Resources. A. Kouvatsi: Project administration. A.G. Ramsay: Resources, supervision, writing-original draft, writing-review and editing. K. Stamatopoulos: Conceptualization, resources, supervision, funding acquisition, writing-original draft, writing-review and editing. A. Chatzidimitriou: Supervision, funding acquisition, project administration.
This work was supported in part by ERANET Transcan-2, Project Acronym: Novel, and the Hellenic Precision Medicine Network in Oncology. A. Vardi is recipient of a research fellowship from the European Hematology Association (EHA Physician Scientist Research Grant). The authors thank Theodoros Moysiadis for providing statistical assistance and Tsagiopoulou Maria for providing assistance with fishplot visualizations.
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.>