A single bout of vigorous intensity exercise enhances the efficacy of rituximab against human chronic lymphocytic leukaemia B-cells ex vivo

which relies on natural killer (NK)-cells binding to rituximab-bound CD20 on B-cells. CLL cells are diffusely spread across lymphoid and other bodily tissues, and ADCC resistance in survival niches may be due to several factors including low NK-cell frequency and a suppressive stromal environment that promotes CLL cell survival. It is well established that exercise bouts induce a transient relocation of NK-cells and B-cells into peripheral blood, which could be harnessed to enhance the efficacy of rituximab in CLL by relocating both target and effector cells together with rituximab in blood. In this pilot study, n = 20 patients with treatment-naïve CLL completed a bout of cycling 15 % above anaerobic threshold for ~ 30-minutes, with blood samples collected pre-, immediately post-, and 1-hour post-exercise. Flow cytometry revealed that exercise evoked a 254 % increase in effector (CD3 (cid:0) CD56 + CD16 + ) NK-cells in blood, and a 67 % increase in CD5 + CD19 + CD20 + CLL cells in blood (all p < 0.005). NK-cells were isolated from blood samples pre-, and immediately post-exercise and incubated with primary isolated CLL cells with or without the presence of rituximab to determine specific lysis using a calcein-release assay. Rituximab-mediated cell lysis increased by 129 % following exercise ( p < 0.001). Direct NK-cell lysis of CLL cells – independent of rituximab – was unchanged following exercise ( p = 0.25). We conclude that exercise improved the efficacy of rituximab-mediated ADCC against autologous CLL cells ex vivo and propose that exercise should be explored as a means of enhancing clinical responses in patients receiving anti-CD20 immunotherapy.


Introduction
Chronic lymphocytic leukaemia (CLL) is the second most common adult blood cancer in the UK (HMRN, 2019), with incidence peaking among people aged ≥ 65 years (Smith et al., 2018).CLL is characterised by the clonal proliferation of mature B-cells in the blood, bone marrow, lymph nodes and spleen (Hallek et al., 2018).Often diagnosed incidentally, the clinical course of CLL is variable, ranging from indolent to aggressive.Patients typically present with asymptomatic, early-stage disease at diagnosis (Hallek et al., 2018), and are monitored without therapy until progression to symptomatic CLL.These patients are referred to as treatment-naïve (Hallek et al., 2018).Once disease progression occurs, relatively more physically fit, younger patients have, for nearly two decades, typically received anti-CD20 monoclonal antibody (mAb) therapy (e.g.rituximab) combined with chemotherapy agents fludarabine and cyclophosphamide (FCR) (Eichhorst et al., 2016;Hallek, 2019;Hallek et al., 2010).Despite inducing favourable responses, rituximab with FC often fails to induce disease eradication.Several potential reasons have been described, related to both tumour cell intrinsic and extrinsic factors (Marshall et al., 2017), and include the failure of antibody-dependent cellular cytotoxicity (ADCC).
Rituximab was one of the first mAb therapies approved for use in clinical practice and eliminates cancer cells via (i) antibody-dependent cellular cytotoxicity, (ii) complement-dependent cytotoxicity, (iii) antibody-dependent cellular phagocytosis, and (iv) direct apoptosis (Manches et al., 2003).During an ADCC response, rituximab binds to CD20 on the CLL cell surface, allowing interactions between Fcγ receptors (such as CD16a) on NK-cells and monocytes, and the Fc portion of opsonised target cells, resulting in cell lysis via exocytosis of perforins and granzymes (Iannello and Ahmad, 2005).Rituximab with chemotherapy (e.g., FCR) achieves high complete remission rates, but, it does not induce disease eradication (Hallek et al., 2010;Fischer et al., 2016).CLL cells are typically diffusely spread across lymphoid and other bodily tissues (Kipps et al., 2017;Thompson et al., 2018;Böttcher et al., 2012) and ADCC resistance in survival niches is thought to be driven by a variety of factors including a suppressive stromal environment that promotes CLL cell survival (Burger et al., 2009) and low CD56 dim NKcellswhich express CD16 (Morvan and Lanier, 2016) frequency in lymphoid tissue (De Weerdt et al., 2019).CLL cells that survive treatmentknown as minimal residual disease (MRD) -exist below the detection threshold used for complete remission determination via flow cytometry (<1 CLL cell per 10 4 leukocytes) (Hallek et al., 2018;Rawstron et al., 2016).Therefore, disease relapse is inevitable for most patients, and is often aggressive and treatment resistant (Cramer et al., 2015;Fink et al., 2013).Thus, novel approaches to tackle MRD in the setting of rituximab are warranted.
It is well established that during an individual bout of moderate to vigorous intensity aerobic exercise in healthy humans, immune cell frequency in peripheral blood increases profoundly.Driven by increased haemodynamic forces and catecholamine release in blood (Shephard, 2003;Simpson et al., 2015), the preferential, transient relocation of immune cell subtypes into the bloodstream is dependent on the cells expression of β 2 -adrenergic receptors and their activation in response to increased adrenaline (Simpson et al., 2015;Graff et al., 2018).For example, NK-cells with a cytotoxic phenotype (CD3 − CD56 dim ) are the most responsive, increasing by up to 1000 % (Campbell et al., 2009;Bigley et al., 2014), followed by CD8 + T-cells (+140 %) (Campbell et al., 2009), and CD19 + B-cells (+130 %) (Turner et al., 2016).Given that CD3 − CD56 + CD16 + NK-cells are required for rituximab-mediated ADCC, it may be that exercise is a potent means of enhancing the efficacy of rituximab against CLL cells in blood.Additionally, it has been shown in a study of n = 16 trained cyclists, that NK-cells expressing the terminal differentiation marker (CD57) increase in the blood by up to 500 % following 30-minutes of cycling exercise (15 % above lactate threshold) (Bigley et al., 2014).CD57 + NK-cells contain large amounts of granzymes and perforins (Morvan and Lanier, 2016;Lopez-Vergès et al., 2010), and may be more sensitive to CD16 activation (Lopez-Vergès et al., 2010) than their immature counterparts and therefore, this phenotype of preferentially mobilised cells during exercise may enhance ADCC responses further.Importantly, cycling for 20-minutes at 80 % maximal oxygen uptake (V˙O 2MAX ) has been shown to induce an 88 % increase in blood CD19 + B-cells in n = 9 healthy males (Turner et al., 2016).Further, human B1 cells (CD19, CD27, CD43) which correspond to the phenotype of CLL cells (Kikushige, 2020), were mobilised by 84 % in response to exercise in the same study (Turner et al., 2016).It may therefore be the case that exercise can be harnessed to mobilise CLL cells into the blood, where they may be more susceptible to the mechanismsof-action of rituximab.A rodent model of acute stresswhich induces a similar adrenergic response to individual bouts of exerciseappeared to relocate B-cells from the bone marrow into the circulation (Engler et al., 2004) suggesting that B-cells mobilised in response to exercise may have lymphoid origins.The redistribution of CLL cells from the protection of the lymphoid tissues, into blood, may support efforts to delete more treatment resistant cells such as those responsible for MRD in CLL.Taken together, we hypothesise that an individual bout of cycling exercise will mobilise CLL cells from diffuse lymphoid tissues into the peripheral circulation, and will concomitantly mobilise cytotoxic NK-cells into blood, which in turn may enhance the efficacy of rituximab.
It is currently unknown how an individual bout of exercise influences circulating immune cell frequency in people with CLL.Therefore, the aim of this study was to characterise the frequency of immune cells and CLL cells in blood of people with treatment-naïve CLL in response to an individual bout of vigorous intensity cycling exercise, and determine if immune cell changes in blood correspond with improvements to rituximab efficacy ex vivo against autologous CLL cells.

Participants
Seventy-three patients diagnosed with CLL and aged ≥ 18-years were invited to participate in the study.Diagnosis of CLL was in-line with the 2018 International Workshop on CLL (iwCLL) guidelines, based on the presence of 5 × 10 3 B-cells/μL blood, sustained for ≥ 3-months and confirmed by blood smear, immunophenotyping and/or epigenetics (Hallek et al., 2018).
Fifty-four treatment-naïve CLL patients were approached about the trial; treatment-naïve CLL was defined as asymptomatic, early-stage disease which is routinely monitored without treatment.Of n = 54 treatment-naïve CLL patients approached, n = 26 declined to participate in the trial.Initial telephone screening involving verbal completion of the Physical Activity Readiness Questionnaire (Thomas et al., 1992) resulted in n = 2 patients being excluded (n = 1 due to dyspnoea, n = 1 due to atrial fibrillation).A further n = 2 patients were excluded prior to the screening visit as the target sample size had been achieved.As such, n = 24 (44 %) treatment-naïve CLL patients attended the screening visit (Fig. 1).During the screening visit, patients were assessed for exclusion criteria: severe/uncontrolled cardiovascular, neurological, psychiatric, metabolic, and respiratory disease, musculoskeletal conditions that would impair the ability to cycle, metastatic/palliative malignancy, abnormal haematological results (e.g., haemoglobin < 100 g/dL), and an Eastern Cooperative Oncology Group (Oken et al., 1982) performance status > 1.A 12-lead resting electrocardiogram (ECG) was recorded and reviewed by a cardiologist.Abnormal resting ECG resulted in n = 2 patients being excluded, meaning n = 22 patients with treatment-naïve CLL were recruited.
Our study was primarily focussed on patients with treatment-naïve CLL, and their results are discussed herein, unless otherwise stated.However, to explore whether cycling exercise could mobilise MRD CLL cells during remission, a subset of CLL patients in remission were also recruited.To be eligible, remission patients had achieved complete remission with incomplete marrow recovery or partial remission for ≥ 6months following the completion of anti-CD20 treatment, as per iwCLL guidelines (Hallek et al., 2018).Inclusion/exclusion criteria were as described for treatment-naïve CLL.A total of n = 19 remission patients were approached.Two patients failed telephone screening (n = 2 due to abnormal haematological results) and nine patients declined to be involved in the trial.Screening of n = 8 remission patients resulted in n = 4 being excluded due to abnormal resting ECG, and n = 4 being recruited, with n = 1 withdrawing from the study before completion due to illness unrelated to the trial (Fig. 1).All participants provided written informed consent prior to completing pre-experimental and experimental procedures.This study was approved by the National Health Service West of Scotland Research Ethics Committee 4 (20/WS/0049) and registered on ClinicalTrials.gov(NCT05093192).

Pre-experimental procedures
Participants were instructed to avoid strenuous exercise for 24-hours prior to the visit and arrived following a ≥ 4-hour fast.Height was assessed by a stadiometer (Seca, Birmingham, United Kingdom), body mass and fat percentage were determined by electronic scales and bioelectrical impedance analysis (Tanita InnerScan, Tokyo, Japan).Participants rested in a supine position for ~ 10-minutes prior to three blood pressure measurements using an automated blood pressure monitor (SunTech TangoM2, Wuxi, China).Anaerobic threshold was determined by a submaximal ramp incremental exercise test on a cycle ergometer (Lode Excalibur, Groningen, The Netherlands).Participants were instructed to maintain a cadence between 60 and 80 revolutions per minute throughout, and completed a 3-minute warm-up of unloaded cycling, followed by a 5-25 W⋅min − 1 incremental phase up to a rating of perceived exertion (RPE) of ≥ 17 on a Borg 6-20 scale (Borg, 1982), or until the researchers were satisfied that anaerobic threshold was achieved (e.g., respiratory exchange ratio > 1.0), and then an unloaded cool-down for 5-minutes.Breath-by-breath gas exchange/ventilation (Carefusion Vyntus CPX, CA, USA), heart rate via 12-lead ECG (Carefusion Vyntus ECG, CA, USA), and arterial oxygen saturation (SpO 2 ) via pulse oximetry (Nonin PureSAT, MN, USA) were recorded continuously during exercise, whilst RPE was recorded every 60-seconds and blood pressure was recorded pre-and immediately post-incremental exercise.Pulmonary oxygen uptake (V˙O 2 ), carbon dioxide production (V˙CO 2 ), and ventilatory equivalents of O 2 (V˙E/V˙O 2 ) and CO 2 (V˙E/V˙CO 2 ) data were interpolated to 15-second averages.The V-slope method (Beaver et al., 1986) was used to determine anaerobic threshold and was further confirmed through visual inspection of V˙E/V˙O 2 and V˙E/V˙CO 2 measurements, independently by two researchers.Anaerobic threshold was reported in terms of V˙O 2 (mL⋅kg − 1 ⋅min − 1 ), power output (W), and as a percentage of age predicted maximum heart rate, using the following equation: % Age predicted heart rate max = (measured heart rate ÷ (220 -age in years)) × 100.

Experimental procedures
At least 3-days after pre-experimental procedures, participants attended the experimental visit between 07:00-10:00 having avoided strenuous exercise and alcohol for 24-hours, and caffeine for ≥ 9-hours, and having fasted for ≥ 9-hours.Body mass was reassessed, and participants rested in a supine position for ~ 30-minutes prior to three blood pressure measurements and a 50 mL blood sample drawn from the antecubital fossa by venepuncture (baseline).The exercise intervention consisted of a 5-minute warm-up at 10 % of the subsequent workload, and 20-to 30-minutes of cycling at a workload corresponding to 10-15 % above individual anaerobic threshold, whilst maintaining a cadence of 60-80 revolutions per minute throughout.Breath-by-breath gas exchange/ventilation, heart rate, and SpO 2 data were recorded continuously during exercise and interpolated to 5-minute averages, whilst RPE was recorded every 5-minutes.Following exercise cessation, participants were immediately positioned in a supine position and a 50 mL blood sample was drawn within 3-minutes (post-exercise) and a final 20 mL blood sample was drawn after 1-hour (1-hr post-exercise).
EDTA-treated whole blood was analysed for blood lactate and blood glucose concentrations using rapid analysers (Lactate Plus Meter, Nova Biomedical, MA, USA and FreeStyle Optium Neo, Berkshire, UK, respectively).Subsequently, plasma was prepared from EDTA-treated blood by centrifugation at 2,000 × g for 15-minutes at 4℃ within 4hours of collection.Serum was prepared from silica-treated blood, which was allowed to clot for ≥ 60-minutes, prior to centrifugation at 1,300 × g for 10-minutes at 4℃.Both plasma and serum were cryopreserved at − 80℃ for long-term storage.

Leukocyte differential
EDTA-treated whole blood was analysed in triplicate using an automated haematology analyser (Sysmex KX-21 N, Kobe, Japan) for leukocyte and erythrocyte frequencies, haemoglobin, haematocrit, platelets, and proportions/numbers of lymphocytes, monocytes, neutrophils, eosinophils and basophils.In n = 9 treatment-naïve CLL patients, lymphocyte and monocyte proportions/numbers could not be determined by an automated haematology analyser.In these instances, monocyte proportions were determined by flow cytometry as HLA-DR + cells in the monocyte domain of a forward scatter-area × side scatterarea gate and lymphocyte proportions/numbers were calculated by negative deduction thereafter (Supplementary Fig. S1).
All antibodies were pre-titrated to ensure optimal fluorescent staining was achieved (data not shown).Both unstained cells from each timepoint, and single stained tubes containing anti-mouse positive (Igκ) and negative control compensation particles (BD Biosciences, Wokingham, UK) were used in each assay to account for spectral overlap.Data were analysed in FlowJo (Version 10.9, BD Biosciences, Wokingham, UK) according to the representative gating strategies reported in Figs. 2 and  3.

Antibody-dependent cellular cytotoxicity (ADCC) assay
CLL cells were isolated from 1 mL of sodium heparin-treated whole blood collected pre-exercise by EasySep™ direct immuno-magnetic negative selection, as per manufacturer's instruction (StemCell Technologies, Vancouver, Canada), achieving a mean purity of 87 ± 15 % viable CD19 + CD56 − cells (Supplementary Fig. S2).Enriched CLL cells were resuspended in 2 mL PBS and labelled with 10 µg/mL calcein acetoxymethyl ester (calcein-AM) (Invitrogen TM , Thermo Fisher Scientific, Loughborough, UK).Non-fluorescent calcein-AM passively diffuses across the membrane of live cells, where it is converted to green, fluorescent calcein following acetoxymethyl ester hydrolysis by intracellular esterases, which renders it membrane-impermeable.Following removal of excess calcein-AM through washing, the relative fluorescence intensity in the supernatant is proportional to cell lysis.
Labelled CLL cells were washed thrice by centrifugation at 500 × g for 5-minutes at room-temperature, and resuspended in 15 mL PBS, followed by a final resuspension at 2 × 10 5 cells/mL in PBS.CLL cells were seeded at a concentration of 5 × 10 3 /well in U-shaped, non-tissue culture treated 96-well plates (Falcon®, Corning, NY, USA) and experimental wells were immediately treated, for 30-minutes at 37℃ and 5 % CO 2 , with 10 µg/mL of either anti-CD20 rituximab (ADCC) (Selleckchem, TX, USA) or anti-HER2 herceptin (antibody-independent cellular cytotoxicity [AICC]) (isotype control; Selleckchem, TX, USA).CLL cells were also seeded in control wells, which were cultured in PBS (10 % [v/v] HI-FCS) to determine spontaneous lysis (negative control) and maximum lysis (positive control).NK-cells from each time-point were isolated by EasySep™ immunomagnetic negative selection from PBMCs, as per manufacturer's instruction (StemCell Technologies, Vancouver, Canada), achieving a mean purity of 68 ± 30 % CD19 − CD56 + cells pre-exercise, and 80 ± 24 % CD19 − CD56 + cells post-exercise (Supplementary Fig. S2).Enriched NK-cells were suspended in PBS at 2 × concentration of the blood they were isolated from and, subsequently, diluted 1:1 in the 96-well plate to establish an NK-cell concentration representative of blood.To investigate whether CLL cell lysis was specific to the interaction of target cell bound rituximab with CD16 on NK-cells (i.e., ADCC), a fraction of NKcells were treated with 50 µg/mL anti-CD16 mAb (B73.1)(ADCC + anti-CD16) (Invitrogen TM , Thermo Fisher Scientific, Loughborough, UK) for 1-hour at room-temperature, prior to being added to the 96-well plate.It should be noted here that two isoforms of CD16 exist (CD16a, CD16b) with CD16a considered the primary receptor involving NK-cell mediated ADCC.Nevertheless the CD16 on NK-cells is typically the CD16a isoform, with CD16b restricted neutrophils (Ravetch and Perussia, 1989).Culture was performed in 10 % (v/v) HI-FCS.In a subset of patients, the influence of human plasma on ADCC was explored, where culture of NK-cells isolated pre-exercise and immediately postexercise was performed in 10 % (v/v) plasma collected pre-exercise, Fig. 2. Representative gating strategy for CLL cells from one participant with treatment-naïve CLL.A) Doublets and debris were removed by plotting FSC-H against FSC-A, lymphocytes were then gated from singlets by plotting SSC-A against FSC-A.Using a fixable viability stain (FVS) viable cells were gated as FVS620 − using a histogram.CD3 was then gated in viable cells to remove any T cells from further analysis using a histogram.B) In a pre-exercise sample, following the exclusion of T cells, CD19 was plotted against CD5 to identify total CLL cells (CD5 + CD19 + ).In the CLL cell population, FSC-A was plotted against sIgM to identify B-cell receptor engagement, and FSC-A was plotted against CD20 to identify CLL cells susceptible to rituximab.CLL cells susceptible to rituximab (CD5 + CD19 + CD20 + ) were then gated to identify subsets with prognostic relevance (FSC-A against CD49d and FSC-A against CD38), a phenotype consistent with a propensity to migrate (CD5 dim CXCR4 bright ), and with a phenotype consistent with recent egress from lymphoid tissue (CD5 bright CXCR4 dim ).C) Represents the same gating strategy as in 'B)' but from a post-exercise sample.CLL, chronic lymphocytic leukaemia; FSC-H, forward scatter-height; FSC-A, forward scatter-area; SSC-A, side scatter-area.
or 10 % (v/v) plasma collected post-exercise, respectively.In all conditions, PBS was added to each well to achieve a final volume of 200 µL after addition of CLL cells, NK-cells, mAb, and HI-FCS or human plasma.The plate was incubated for 2-hours at 37℃ and 5 % CO 2 .
Following the 2-hour incubation, maximum lysis wells were treated with 100 µL/well of 4 % (v/v) Triton X-100 (Invitrogen TM , Thermo Fisher Scientific, Loughborough, UK) and the plate was centrifuged at 100 × g for 2-minutes at room temperature.Subsequently, 75 µL/well of acellular supernatant was transferred to a 96-well flat-bottom black plate (Corning TM , Thermo Fischer Scientific, Loughborough, UK).Fluorescence (485 nm, 530 nm) was measured using a Pherostar plate reader (BMG Labtech, Ortenberg, Germany) with the gain (based on positive controls) and read height optimised for each plate.Arbitrary relative fluorescence was converted to the percentage of specific lysis using the following equation: Spontaneous and maximum lysis conditions were evaluated over six replicates, whilst experimental wells were seeded in at least triplicate.All ADCC assay procedures were conducted in aseptic conditions.Assay procedures for remissive CLL patients are described in Supplementary Methods.

Statistical analysis
Statistical analyses were performed using SPSS (IBM SPSS Statistics Version 28, IL, USA).Data are presented as mean ± standard deviation (SD) unless otherwise stated.Paired sample t-tests or Wilcoxon signedrank tests (if non-parametric distribution was observed following a Shapiro-Wilk test) were used to compare differences pre-to postexercise for ADCC, AICC, and ADCC + anti-CD16.Statistical analysis was one-way repeated measures analysis of variance (ANOVA) for Fig. 3. Representative gating strategy for NK-cells from one treatment-naïve CLL participant.A) Doublets and debris were removed by plotting FSC-H against FSC-A, lymphocytes were then gated from singlets by plotting SSC-A against FSC-A.Using a fixable viability stain (FVS) viable cells were gated as FVS620 − using a histogram.CD14 was then gated in viable cells to remove any monocytes from further analysis using a histogram.B) In a pre-exercise sample, following the exclusion of monocytes, CD56 was plotted against CD3 to isolate total NK-cells (CD3 − CD56 + ).In the CD3 − CD56 + population, CD56 dim and CD56 bright NK-cells were gated by plotting FSC-H against CD56, CD16 − and CD16 + NK-cells were gated using a histogram, and CD57 − and CD56 + NK-cells were gated using a histogram.In both the CD56 dim and CD56 bright populations, NKG2A was plotted against CD158a to determine NK-cell differentiation status including, low-differentiated (CD158a − NKG2A + ), medium-differentiated (CD158 + NKG2A + ) and highly differentiated (CD158a + NKG2A − ).C) Represents the same gating strategy as 'B)' but from a post-exercise sample.CLL, chronic lymphocytic leukaemia; FSC-H, forward scatter-height; FSC-A, forward scatter-area; SSC-A, side scatter-area.

Treatment-naïve CLL
Of n = 22 treatment-naïve CLL patients that provided informed consent for the trial, n = 20 participants completed all experimental procedures, and their baseline characteristics are reported in Table 1.

Safety
One adverse event (syncope) occurred following pre-experimental exercise procedures.No other adverse events or serious adverse events occurred.

Characteristics of cycling exercise
All participants completed a single bout of cycling at a workload corresponding to 10-15 % above anaerobic threshold, and n = 14 (70 %) completed 30-minutes of cycling.The remaining n = 6 (30 %) participants ceased cycling after 21-minutes and 36-seconds ± 3-minutes and 18-seconds (range: 20-minutes and 0-seconds to 28-minutes and 30-seconds) due to volitional exhaustion.Importantly, the n = 6 who did not complete the prescribed 30-minutes exercise duration exhibited physiological responsessuch as RPE and blood lactatein line with the rest of the cohort (Supplementary Tables S2 and S3) and therefore were included in subsequent within group analyses.Physiological responses including, V˙O 2 (mL⋅kg − 1 ⋅min − 1 ), V˙O 2 relative to anaerobic threshold (%), V˙CO 2 (mL⋅kg − 1 ⋅min − 1 ), V˙E (L⋅min − 1 ), respiratory exchange ratio, heart rate (bpm), heart rate as a percentage of age-predicted maximum (%), and RPE (6-20 Borg scale) (Borg, 1982) were averaged into two, 15-minute segments and are displayed in Table 2. Participants cycled above their anaerobic threshold throughout the cycling trial, with a significant increase in V˙O 2 relative to anaerobic threshold from warm-up to 15-and 30-minutes (61 ± 18 % vs 116 ± 30 %, p < 0.001 and 127 ± 31 %, p < 0.001) in addition to a significant increase from 15-minutes to 30-minutes (p < 0.001).Blood lactate increased by 145 % (p < 0.001) from pre-to post-exercise (Table 3).The effect of the acute bout of cycling on leukocyte, lymphocyte, monocyte, neutrophil, and erythrocyte frequency are displayed in Table 3.

ADCC changes in response to cycling exercise
There was a significant 129 % increase observed in ADCC pre-to post-exercise (t (17) = 5.04, p < 0.001, d = 1.19) with no differences preto post-exercise in AICC (Z = 1.61, p = 0.11) or ADCC + anti-CD16 (Z = 0.94, p = 0.35) on CLL cell lysis (Fig. 4A).We note that a significant 12 % difference was observed between the purity of NK-cells isolated preexercise and post-exercise (Z = 2.36, p = 0.018).As a result, we determined whether this difference in NK-cell purity accounted for the change in ADCC responses observed, but it did not alter the main effects observed (see Supplementary Results).

NK-cell mobilisation during cycling exercise
There was a significant effect of time on the number of CD3 Indicates a significant difference from 'warm-up' at p < 0.001, † † † indicates a significant difference from '15-minutes' at p < 0.001, following post hoc Bonferroni comparisons.CLL, chronic lymphocytic leukaemia; ANOVA, analysis of variance.

Table 3
Haemodynamic variables pre-exercise, immediately post-exercise, and 1-hr post-exercise with percentage change (%Δ) pre-to post-exercise and main effect from one-way repeated measures ANOVA in treatment-naïve CLL.Data are mean ± SD, n = 20.** Indicates a significant difference from pre-exercise at p < 0.01, *** indicates a significant difference from pre-exercise at p < 0.001, † indicates a significant difference from post-exercise at p < 0.05, † † indicates a significant difference from post-exercise at p < 0.01, † † † indicates a significant difference from postexercise at p < 0.001 following post hoc Bonferroni comparisons.ANOVA, analysis of variance; CLL, chronic lymphocytic leukaemia.We also explored whether an individual bout of vigorous intensity exercise could mobilise CLL cells into the blood in a subset of CLL patients in remission.All remission patients included (n = 3) were MRD negative in blooddefined as < 1 CLL cell/10 4 leukocyteswith no change observed in the number of total CD5 + CD19 + CLL cells pre-to post-exercise (F (2,4) = 1.23, p = 0.38, ηp 2 = 0.38) (Supplementary Table S5).

Table 4
Treatment-naïve CLL NK-cell subsets pre-exercise, immediately post-exercise, and 1-hr post-exercise with percentage change (%Δ) pre-to post-exercise, and main effect from repeated measures ANOVA.Data are mean ± SD, n = 19.
*Indicates a significant difference from pre-exercise at p < 0.05, ** indicates a significant difference from pre-exercise at p < 0.01, *** indicates a significant difference from pre-exercise at p < 0.001, † indicates a significant difference from post-exercise at p < 0.05, † † indicates a significant difference from post-exercise at p < 0.01, † † † indicates a significant difference from post-exercise at p < 0.001 following post hoc Bonferroni comparisons.ANOVA, analysis of variance.

Table 5
Treatment-naïve CLL B-cell subsets pre-exercise, immediately post-exercise, and 1-hr post-exercise, with percentage change (%Δ) pre-to post-exercise and main effect from repeated measures ANOVA.Data are mean ± SD, n = 20.F (1.10,20.96)= 15.45,p < 0.001, ηp 2 = 0.45 *Indicates a significant difference from pre-exercise at p < 0.05, ** indicates a significant difference from pre-exercise at p < 0.01, † indicates a significant difference from post-exercise at p < 0.05, † † indicates a significant difference from post-exercise at p < 0.01, † † † indicates a significant difference from post-exercise at p < 0.001 following post hoc Bonferroni comparisons.CLL, chronic lymphocytic leukaemia; ANOVA, analysis of variance.

Discussion
In this study, we analysed the effects of an individual bout of vigorous intensity cycling on the frequency of immune cells and CLL cells in the blood of people with treatment-naïve CLL, and investigated whether immune cell changes in blood correspond to improvements in rituximab-mediated ADCC against autologous CLL cells ex vivo.NK-cells and their subsets increased immediately post-exercise.ADCC increased post-exercise when autologous NK-cells and CLL cells were incubated in the presence of rituximab.Additionally, CLL cells also increased during exercise, including those with a recently egressed from lymphoid tissue phenotype.
This study is the first to show that a single bout of vigorous intensity cycling exercise improves rituximab-mediated ADCC against autologous CLL cells ex vivo.The effects of ADCC were blunted when NK-cells expressing CD16 -which make up approximately 55-75 % of total CD3 − CD56 + NK-cells hereinwere blocked with an anti-CD16 antibody.Additionally, exercise had no effect on AICC against autologous CLL cells, which is unsurprising given that CLL cells in treatment-naïve CLL patients are poorly immunogenic to NK-cells, and solely mobilising immune cells via exercise is unlikely to elicit anti-tumour effects if those cells lack immunogenicity (Emery et al., 2022).Taken together these findings indicate that the increase in ADCC was independent of a change in AICC and suggests that the increase was dependent on the preferential mobilisation of NK-cells, specifically the mobilisation of CD3 − CD56 + CD16 + NK-cells (+254 %) -supported by the positive correlation between CD3 − CD56 + CD16 + NK-cells and ADCC responses.It is important to note that the purity of isolated NK-cells used in ADCC assays was 12 % greater in samples collected immediately post-exercise, likely due to NK-cells increasing in proportion (of total lymphocytes) post-exercise (1.6 %) compared to pre-exercise (1.0 %).Nevertheless, controlling for purity as a confounding variable did not nullify the ADCC responses observed.As NK-cells differentiate, their cytotoxic potential increases (Béziat et al., 2010).Terminally differentiated NK-cells (CD57 + ) contain granzymes and perforins (Morvan and Lanier, 2016;Lopez-Vergès et al., 2010) important for ADCC mediated elimination of target cells (Iannello and Ahmad, 2005) and are more sensitive to CD16 activation compared to their immature counterparts (Lopez-Vergès et al., 2010).We also showed that cycling exercise induced a preferential mobilisation of mature, cytotoxic CD3 − CD56 + CD57 + CD16 + NK-cells (+323 %), which also positively correlated with ADCC responses.The mobilisation of mature, cytotoxic NK-cells is consistent with previous findings showing an increase of ~ 500 % in CD3 − CD56 + CD57 + NK-cells following 30-minutes of cycling 15 % above lactate threshold (analogous to anaerobic threshold) (Bigley et al., 2014).Indeed, the mobilisation of mature, cytotoxic NK-cells herein was ~ 180 % less than previously reported.However, the previous study did not include CD16 in their NK-cell phenotyping and recruited healthy, trained cyclists resulting in a greater relative exercise intensity (age predicted heart rate max = 88 ± 4 %) (Bigley et al., 2014) compared to our study (age predicted heart rate max = 75 ± 9 %).
In a subset of participants, we also assessed rituximab-mediated ADCC when target cells were cultured with autologous time-point matched (i.e., pre-and post-exercise) plasma.We did this to explore a more physiologically relevant milieu containing autologous NK-cells, Bcells and plasma.ADCC with plasma increased by 92 % pre-to postexercise, and a medium effect was observed for greater ADCC with plasma compared to ADCC HI-FCS pre-and post-exercise (d = 0.61 and d = 0.70, respectively), with a specific trend towards a difference postexercise (p = 0.069).This medium effect size indicates a meaningful difference, with the non-significant change likely attributed to the smaller sample size (n = 9).The trend towards a difference between ADCC with plasma and ADCC HI-FCS may be explained by the presence of hormones and cytokines (e.g., cortisol and IFN-γ) following vigorous cycling exercise (Gupta et al., 2018).Indeed, it is likely that incubating purified NK-cells with serum collected 1-hour post-exercise could enhance ADCC activity further due to reduced cortisol and elevated IFNγ serum concentrations resulting in enhanced NK-cell activity per cell, as previously reported (Gupta et al., 2018).
Our ADCC assay used NK-cells as effector cells, however, in vivo, other immune cells play a role in the mechanism-of-action of rituximab, including for example, monocytes and macrophages (Sopp and Cragg, 2018).Non-classical (HLA-DR + CD14 dim CD16 + ) monocytes and HLA-DR + CD14 + CD32 + monocytes were elevated by 147 % and 59 % postexercise, respectively, and these cells can initiate ADCC (Vermi et al., 2018).Thus, the mobilisation of effector monocytes may further enhance rituximab's efficacy in vivo.It is important to note here that immunosuppressive MDSCs, were also increased post-exercise (+70 %), and these cells have been shown to inhibit NK-cell mediated ADCC against solid tumour cell lines in vitro, attributed to MDSC secretion of nitric oxide (Stiff et al., 2018).Additionally, the calcein-release assay used herein requires only a small, fixed number of target cells to assess the impact of effector cell changes.As such, in our assay, we fixed the number of target cells at 5 × 10 4 cells/well and those cells were isolated from blood sampled pre-exercise.Given this, we posited that our findings would have greater, clinical relevance if a concomitant increase in CLL cells was also observed following exercise, as discussed below.
The mobilisation of CD5 + CD19 + CLL cells and their subsets in response to cycling exercise in this study is a novel finding not previously described.CLL cells possess a similar phenotype to that of healthy B-cells (e.g., CD19) and these cells may increase up to 100 % following a bout of vigorous cycling in healthy humans (Campbell et al., 2009).Specifically, CLL cells increased by 67 % herein and these cells are broadly consistent with the 'B1′ cell phenotype (Rawstron et al., 2016) which increase by 84 % in response to vigorous cycling exercise in healthy humans (Turner et al., 2016).The difference in magnitude of change could be the result of the different relative exercise intensity in the present study (age predicted heart rate max = 75 ± 9 %) compared to that reported previously (age predicted heart rate max = 94 ± 4 %) (Turner et al., 2016) as exercise-induced lymphocytosis is intensitydependent (Campbell et al., 2009;Bigley et al., 2014).We also assessed whether CLL cells mobilised by cycling exercise expressed CD20 -rendering those cells susceptible to rituximab-mediated ADCC in vivoand we observed a 67 % increase to CD5 + CD19 + CD20 + CLL cells with no change to the expression of surface CD20.Additionally, cycling exercise appeared to relocate CLL cells from their protective stromal niches in lymphoid tissue into the bloodtypified as CD20 + CD5 bright CXCR4 dim (Calissano et al., 2011;Pasikowska et al., 2016) and increased the frequency of CLL cells with prognostic relevance (CD20 + CD49d + ; +72 % and CD20 + CD38 + , +66 %).In a subset of remissive CLL patients we explored whether a single bout of vigorous intensity cycling could mobilise CLL cells into the blood.All remissive patients included in the study were MRD negative in blooddefined as < 1 CLL cell/10,000 leukocyteswith no change in the number of CLL cells observed during vigorous intensity cycling (Supplementary Table S5).
The findings of our study have potentially important clinical implications.For instance, rituximab is effective in achieving complete remission rates in many patients, but it does not induce disease eradication (Hallek et al., 2010;Fischer et al., 2016).This is because CLL cells are typically diffusely spread across lymphoid and other bodily tissues (Kipps et al., 2017;Thompson et al., 2018;Böttcher et al., 2012) and ADCC resistance in survival niches is thought to be driven by a variety of factors including a suppressive stromal environment that promotes CLL cell survival (Burger et al., 2009) and an inadequate frequency of CD56 dim NK-cells expressing CD16 (Morvan and Lanier, 2016) in lymphoid tissue (De Weerdt et al., 2019).Thus, relocating CLL cells from their protective stromal niches into the blood may render these cells more susceptible to the mechanisms-of-action of rituximab.This relocation also suggests that individual bouts of vigorous intensity exercise may have the potential to be employed alongside Bruton's Tyrosine Kinase (BTK) inhibitors, which have been shown to transiently increase lymphocyte frequency, peaking in the blood approximately 4-weeks (+~300 %) after therapy in people with CLL (Byrd et al., 2013;Byrd et al., 2020).The mobilisation of CLL cells also has important implications for patients in CLL remission, specifically, individual bouts of cycling exercise may aid in identifying MRD positivity and may improve the responses to rituximab in a relapse setting.However, preliminary evidence from herein (n = 3) suggests that this may not occur in patients who are MRD negative in blood.Furthermore, the concomitant increase in CD16 + NK-cells and our demonstration of enhanced rituximabmediated ADCC provides a compelling rationale for future research to investigate the effects of individual bouts of exercise alongside rituximab therapy in vivo.Given that rituximab has a half-life of ~ 28-days, which is typically the time when an additional treatment cycle commences (Li et al., 2012), it is not necessary for exercise bouts to be performed during rituximab infusions, but rather, exercise bouts can be performed by patients away from the clinic in the days and weeks after infusion (Collier-Bain et al., 2023).Additionally, a subset of treatmentnaïve patients herein did not complete the 30-minute prescribed cycling bout and yet were observed to exhibit increased immune cell frequency in blood and enhanced rituximab-mediated ADCC.For example, ADCC pre-and immediately post-exercise was 2.8 ± 8.2 % and 11.9 ± 16.7 %, respectively, and CD3 − CD56 + CD16 + NK-cell frequency pre-and immediately post-exercise was 94 ± 71 cells/µL and 444 ± 439 cells/µL, respectively (Supplementary Table S3), suggesting that vigorous intensity exercise for ~ 20-minutes may be sufficient to induce relevant immunomodulation.Collectively, the results of our study suggest that other mAbs used to treat CLL and other haematological cancersparticularly those with neoplastic fractions residing in treatmentresistant nichesmay also benefit from individual bouts of vigorous intensity exercise, as recently described (Collier-Bain et al., 2023).
A strength of this study, but a point that could also be considered a limitation, was the inclusion of treatment-naïve participants.This population enabled the examination of the effects of an individual bout of cycling exercise without confounding variables such as anti-CLL therapieswhich not only would have interfered in our assays, but may cause leukocytopenia, and anaemia (Hallek et al., 2010) or disease associated comorbidities, which together may result in greater contraindications to exercise, such as, cardiac arrhythmias (Hallek and Al-Sawaf, 2021).However, this may be considered a limitation as the effects of exercise on rituximab efficacy was not tested in patients receiving rituximab who may present with greater immune dysregulation such as reduced cytotoxic potential and altered phenotype of NKcells (Parry et al., 2016).Therefore, it is important for future research to assess the safety, and acceptability of moderate to vigorous intensity exercise in patients undergoing anti-CLL therapy, particularly given that chemo-immunotherapy may result in immunosuppression, which may nullify the immunomodulatory changes that arise in response to individual bouts of exercise.
To summarise, our findings show that an individual bout of vigorous intensity cycling temporarily increased the frequency of CD3 − CD56 + CD16 + NK-cells and CD5 + CD19 + CD20 + CLL cells in the blood of patients with treatment-naïve CLL.Our ex vivo investigations demonstrated enhanced rituximab-mediated ADCC because of elevated CD16 + NK-cells in blood.Thus, exercise could be explored as a means of improving clinical responses in patients receiving rituximab, and/or other anti-CD20 monoclonal antibodies such as, obinutuzumab.

Fig. 1 .
Fig. 1.A CONSORT flow diagram of the recruitment, screening and experimental completion of the study.

Table 2
Participant Characteristics.Characteristics of vigorous intensity cycling exercise in treatment-naïve CLL, with main effect from repeated measures ANOVA.Data are mean ± SD, n = 20.