Rapamycin Prevents Surgery-Induced Immune Dysfunction in Patients with Bladder Cancer

Activating genetic alterations in PI3K/AKT/mTOR signaling are present in more than 40% of bladder tumors (1–6), providing justification for using mTOR inhibition to treat bladder cancer if mTOR inhibition is the sole aim. The prototypical mTOR complex 1 (mTORC1) inhibitor, rapamycin, and rapamycin analogues demonstrate effective antitumor activity in preclinical bladder cancer models (7–14) and human trials have been conducted. These trials sought the maximal tolerated dose based on the notion that higher dosing leads to improved inhibition of target tumor cell mTORC1 signaling in the bladder cancer itself. However, at standard clinical doses (i.e., ≥5 mg daily), these agents demonstrate only modest activity against metastatic disease (15–17) and are too toxic for treatment of less advanced disease (11, 16, 17).

In addition to driving tumor growth, mTOR signaling also regulates activation and differentiation of many cell types, including T cells, which mediate antitumor immunity (18–21). We previously showed distinct effects of rapamycin in cancers across dose and tissue types with higher doses (1–8 mg/kg/day) directly suppressing tumor cell growth, whereas lower doses (0.075 mg/kg/day) enhanced T-cell immunity (18). Potential beneficial immune effects of lower dose mTORC1 inhibition are also supported by data from noncancer patients, showing that at 0.5 mg daily or 5 to 20 mg weekly, everolimus, a rapamycin analogue, improves antibody immune responses to an influenza vaccine and reduces the prevalence of programmed death -1 (PD-1)–positive T cells (22). PD-1 is a marker of T-cell exhaustion, a common feature of chronic infections (23) and cancer (24) characterized by ineffective pathogen control or tumor eradication.

To determine the pharmacodynamic and T-cell effects of oral rapamycin in patients with bladder cancer undergoing cystectomy, we conducted a presurgical clinical trial. This approach enabled evaluation of paired patient tissues serially collected during treatment to examine drug delivery and target tissue specificity (25). Comparison of immune cells collected before and after cystectomy revealed a significant increase in T cells expressing PD-1 following cystectomy, suggesting that surgery could inhibit antitumor immunity. We studied clinical potential using the syngeneic MB49 mouse bladder tumor model. Surgery increased lung metastasis and reduced response to PD-L1 antibody immunotherapy in this model. These findings characterize untoward effects of surgery on T-cell function and response to cancer immune therapy and could influence perioperative treatment strategies.

A two-arm, open-label randomized study of rapamycin for patients (n = 20) suffering from invasive urothelial bladder cancer was conducted (trial schema in Supplementary Fig. S1). The local institutional regulatory board (IRB) approved the study (IRB 12-135H), which was publicly registered (ClinicalTrials.gov identifier: NCT01827618). All participating patients provided written informed consent. This study's involvement with human subjects complies with the Declaration of Helsinki. Tissues collected included blood and bladder tumors and were processed for flow cytometry or immunohistochemistry (IHC) as described below. Eligible patients had invasive (clinical stage ≥T1) bladder cancer, with no evidence of nodal or visceral metastasis, Zubrod performance status of 0 to 2, and were unable or unwilling to receive cisplatin-based neoadjuvant chemotherapy. Additional eligibility criteria included adequate bone marrow function (defined as granulocytes >1,500 cells/mm³, hemoglobin ≥9.5 gm/dL, and platelets >70,000/mm³), life expectancy ≥1 year, ability to provide sufficient tumor at surgery for research purposes, and age ≥18 years. Exclusions included hepatic impairment, HIV or other chronic infections, allergy to rapamycin, or chronic steroid use. Patients with controlled medical conditions (e.g., hypertension, diabetes) were eligible if they were under physician care. Enrollment occurred from May 2012 to September 2013. Cystectomy was indicated for all patients with muscle-invasive (stage ≥T2) bladder cancer and for patients with non–muscle-invasive (stage T1) bladder cancer due to inability to clear disease with repeated resections.

Patients were randomized 1:1 to receive 3 mg daily oral rapamycin (Rapamune, Wyeth Pharmaceuticals, 1 mg tablets) irrespective of weight or age, or no treatment for 28 days prior to cystectomy. No dose adjustments were made, but blood concentrations were measured to monitor adherence and to correlate with tissue concentrations. The final dose was given in the morning, one day before surgery. Patients were instructed to take rapamycin 1 hour before or 2 hours after food and to avoid grapefruit juice, which inhibits cytochrome P450 CYP3A4 leading to decreased drug metabolism. Cystectomy included bladder and prostate (males) removal, bilateral pelvic lymph node dissection, and urinary diversion. Wound closure was performed with absorbable sutures (not staples) and drains were left in place until drain flow was acceptably low (≤100 cc per day) as recommended for patients on rapamycin treatment (26).

Toxicity was evaluated using the National Cancer Institute Common Toxicity Criteria version 4.02. Safety and adverse events were assessed at each clinic visit, on the day of cystectomy, every day following cystectomy while the patient was in the hospital and at follow-up visits up until 90 days after surgery. Assessments included documented medical history, physical examination, and laboratory evaluation as appropriate. Dose-limiting toxicity was defined as grade 3 or 4 neutropenia with fever or lasting >7 days, platelets <70,000/mm³, grade ≥3 nonhematologic toxicity, or irreversible grade 2 toxicity related to rapamycin.

Whole blood for rapamycin concentration measurement was collected by peripheral venipuncture into K2-ethylenediaminetetraacetic acid (EDTA) vials (BD, #367844) at 30 days after registration. At radical cystectomy, a small tissue section (approximately 5 mm³) was taken and immediately frozen for rapamycin concentration assessments. Blood rapamycin was quantified according to previously published procedures (27–29) using tandem mass spectrometer for detection. Rapamycin concentration was not performed in one patient who discontinued rapamycin early and who refused rapamycin blood assessment. Rapamycin and ascomycin (ASCO; internal standard) were obtained from LC Laboratories. High-performance liquid chromatography (HPLC) grade methanol and acetonitrile were purchased from Fisher. All other reagents were purchased from Sigma Chemical Company.

For human bladder tumor tissue, 100 mg of calibrator and unknown tissue samples were mixed by sonication (three 5-second bursts) with 10 μL ASCO (0.5 μg/mL, internal standard) and 300 μL of a solution containing 0.1% formic acid and 10 mmol/L ammonium formate dissolved in 95% HPLC grade methanol. After sonication, the samples were vortexed vigorously for 2 minutes, and then centrifuged at 13,000 × g for 5 minutes at 23°C (subsequent centrifugations were performed under the same conditions). Supernatants were transferred to 1.5 mL microfilterfuge tubes, and then 40 μL of the final extracts were injected into the liquid chromatography with tandem mass spectrometry (LC-MS/MS). The ratio of the peak area of rapamycin to that of the internal standard ASCO (response ratio) for each unknown sample was compared against a linear regression of calibrator response ratios at 0, 2, 10, 20, 50, and 100 ng/g to quantify rapamycin. The concentration of rapamycin was expressed as ng/g of tissue.

We used phosphorylation of ribosomal protein S6 (rpS6) at serine 240/244 (assessed by IHC) as an indirect measure of S6 kinase 1 and mTORC1 activity. An H-score was used as a semiquantitative measure of phosphorylation, determined by the percentage of cells scoring positive (0–100) multiplied by staining intensity (0–3; ref. 30). Genitourinary pathologists blinded to treatment groups and sample pairing performed all IHC interpretations. Tissue slides were cut at 5 μm from the entire tumor within formalin-fixed, paraffin-embedded tissue blocks and subsequently probed using antibodies (Cell Signaling Technologies) for phosphorylated (Ser240/244) rpS6 (#2215), rpS6 (#2217), phosphorylated (Ser473) AKT (#4060), AKT (#4691), phosphorylated (Thr37/46) 4E-BP1 (#2855), 4E-BP1 (#9644), and PTEN (phosphatase and tensin homologue deleted on chromosome ten, #9559).

Whole blood was collected by peripheral venipuncture into lithium heparin vials (BD, #367880) at baseline and at 30 and 60 days after registration. We isolated peripheral blood mononuclear cells (PBMC) from patient blood samples using Ficoll-Paque gradients (GE Healthcare). PBMCs were suspended in freezing medium [complete Roswell Park Memorial Institute (RPMI)] with 50% fetal bovine serum and 10% dimethyl sulfoxide (DMSO; Corning, Fisher Scientific) and cryopreserved at –150 °C until analyzed. When sufficient tumor was available at biopsy or cystectomy, a small tumor section (approximately 5 mm³) was excised under sterile conditions and placed into RPMI media containing 1% antibiotic and transported on ice. Tumor tissues were washed with PBS, cut into 1–2 mm pieces, and digested in 5 mL of digestion solution containing 0.05% trypsin, collagenase (1 mg/mL), and DNase (0.25 mg/mL) for 45 minutes at 37°C and processed into single-cell suspensions. PBMCs were thawed in complete RPMI and counted with a Vi-cell XR (Beckman Coulter) before resuspension in flow buffer (2% fetal bovine serum in PBS).

PBMCs (1 × 10⁶) per sample were stained and analyzed using an LSR II flow cytometer and FACSDiva software (BD Bioscience, v6). Validated commercial reagents were used: LIVE/DEAD Fixable Aqua Dead Cell Stain Kit (Life Technologies), fluorochrome-conjugated, monoclonal anti-human CD45 (clone HI30), CD3 (clone HIT3a), CD4 (clone OKT4), CD8 (clone SK1), IFNγ (clone 4S.B3), TNFα (clone MAb11), CD62L (clone DREG-56), CD45RO (clone UCHL1), FoxP3 (clone 206D), PD-1 (clone EH12.2H7), TIM-3 (clone F38-2E2), and LAG-3 (clone 3DS223H eBioscience, Invitrogen; BioLegend). For cytokine staining, cells were stimulated with Leukocyte Activation Cocktail with GolgiPlug (BD Biosciences #550583) for 5 hours according to the manufacturer's protocol. Then they were surface stained, fixed, permeabilized with fixation and permeabilization solution (BD Biosciences, #554722) according to the manufacturer's protocol and then stained for intracellular cytokines.

PBMCs were first labeled with carboxyfluorescein succinimidyl ester (CFSE) and sorted to high purity (>97%) as CD3⁺PD-1⁺ versus CD3⁺PD-1^– T cells using fluorescence-activated cell sorting on a FACSAria II (BD). Then, 0.5 to 1.0 × 10⁶ cells were stimulated in vitro with Dynabeads Human T-Activator CD3/CD28 at a bead to cell ratio of 1:1 (Life Technologies) in 96-well flat-bottom plates. After 3 days, cells were stained and analyzed using flow cytometry to measure CFSE fluorescence signal of both CD4⁺ and CD8⁺ T cells in sorted populations (Supplementary Fig. S2). CFSE in medium control samples of each T-cell subset was used as the baseline for gating proliferated populations. Samples from all patients and time points were run simultaneously using v6 FACSDiva software (BD Bioscience).

Mice were 8 to 12-week-old C57BL/6J males or females (BL6, Jackson Labs). All mice were maintained in specific pathogen-free conditions. All animal studies were performed using procedures approved by the UTHSA Animal Care and Use Program. For experimental bladder cancer metastasis, male mice were challenged intravenously with 5 × 10⁵ MB49 mouse bladder cancer cells. Surgery was a ventral 2–3 cm midline incision through the skin and peritoneum, followed immediately by two-layer closure using 4-0 vicryl suture (Ethicon J214H) to close peritoneum and 7-mm clips to close skin (CellPoint Scientific 203-1000), performed within 15 minutes of tumor challenge. Surgical anesthesia included ketamine HCl (JHP Pharmaceuticals) at 80 mg/kg, xylazine (LLOYD) at 8 mg/kg, and acepromazine maleate (Boehringer Ingelheim Vetmedica) at 1 mg/kg in sterile PBS at a final volume of 300 μL/mouse/20 g body weight. On day 14 after tumor challenge, mice were euthanized, and lungs were harvested in 10% buffered formalin for subsequent H&E staining, and splenocytes were processed for flow cytometry staining using Fixable Viability Dye (eBioscience, Invitrogen), fluorochrome-conjugated monoclonal anti-mouse CD45 (clone 30-F11), CD3 (clone 17A2), PD-1 (clone 29F.1A12), Tim-3 (clone RMT3-23), and Lag-3 (clone C9B7W; BioLegend). For orthotopic bladder cancer experiments, female mice were challenged with 80,000 MB49 cells as described (31) and treated with monoclonal anti-mouse PD-L1 (clone 10F.9G2, Bio X Cell) at 100 μg per mouse in 100 μL by intraperitoneal injection or equivalent isotype control rat IgG2b monoclonal antibody (clone LTF-2, Bio X Cell) on days 7, 12, and 17. Mice were euthanized on day 21 for examination of retroperitoneal tumor-draining lymph nodes (TDLNs) by cytokine ELISPOT assay or followed for survival as noted.

MultiScreen ImmunoSpot plates (Millipore) were coated with anti-mouse IFNγ capture monoclonal antibody (1 μg/mL; clone AN-18, BioLegend) at 4°C. overnight, followed by washes with sterile PBS, then blocked with 1% BSA (Sigma-Aldrich) in sterile PBS at room temperature for 1 hour. TDLN cells (2.5 × 10⁵) in 100 μL complete RPMI (cR-10) medium containing 10% fetal bovine serum (HyClone, GE), 2 mmol/L l-glutamine (Corning, Fisher), 100 IU penicillin, and streptomycin (100 μg/mL; Corning, Fisher) were added into each well. Another 100 μL of cR-10 was added to each well alone (medium control) or plus 2.5 × 10⁵ irradiated (25 Gy, γ-irradiator) MB49 cells. TDLN cells and tumor target cells were mixed and incubated at 37°C in a 5% CO₂ atmosphere for 24 hours. Wells were then washed and incubated with biotin-conjugated anti-mouse IFNγ detection monoclonal antibody (0.25 μg/mL; clone R4-6A2, BioLegend) at 4°C overnight. Wells were then incubated with 1:2,000 streptavidin–AP (KPL) at room temperature for 1.5 hours, followed by washes and developed with BCIP/NBT phosphatase substrate (KPL). The reaction was stopped by deionized water after spots were formed. IFNγ-producing spot-forming cells (SFC) were quantified using an ImmunoSpot analyzer (S6 Micro, Cellular Technology) with ImmunoSpot software, version 6.0 (Cellular Technology). Absolute SFC numbers in response to irradiated MB49 tumor cells were normalized by SFC in medium control for every TDLN sample.

The primary trial objective was to test the PD response of rapamycin defined by a significant inhibition of S6 kinase 1 (S6K1) phosphorylation (comparing posttreatment to baseline values in tissue samples). Secondary trial objectives were to determine rapamycin bladder tissue concentrations and effects of rapamycin on circulating T lymphocytes. The trial sample size was based on the primary objective of identifying a pharmacodynamic (PD) treatment response in bladder tissue, defined as ≥60% inhibition in mTORC1 activity measured indirectly by phosphorylation of its downstream target ribosomal protein (rp) S6 on Ser 240/244 over total rpS6, based on a prior established benchmark (25). The change in mTORC1 activity was estimated using the change in the H-score for phospho (p)-rpS6/total rpS6 in the cystectomy tumor tissue compared with biopsy tumor tissue (25). The endpoint was mTORC1 inhibition (yes, no) in posttreatment tissue. A sample size calculation was based on Fisher exact test of H_o:p_T = p_C versus H₁:p_T≠p_C at α = 0.05, where p_T and p_C are the proportions of treated and control patients experiencing a PD treatment response. Assuming p_C = 0.10 and p_T = 0.60, this study would attain 80% power for testing H_o with n = 17 patients per group. An interim analysis was conducted following enrollment of the first 20 patients due to unexpected increase in rpS6 phosphorylation at the time of cystectomy in control patients. Pathologic tumor downstaging was defined as clinical stage T2 and pathologic stage <T2 or clinical stage >T2 and pathologic stage ≤T2. Pathologic tumor upstaging was defined as clinical stage <T2 and pathologic stage ≥T2 or clinical stage T2 with pathologic stage >T2. A Pearson product–moment correlation coefficient was computed to assess the relationship between the blood and tissue rapamycin concentrations. A Spearman rank-order correlation coefficient was computed to assess the relationship between rapamycin concentrations and PD response due to violation of normality. Nonparametric data were represented as median [interquartile range (IQR)] and parametric data as mean [standard deviation (SD)]. Groups were contrasted on the median or mean with Wilcoxon or t tests as appropriate and on binary outcomes with Fisher exact test. The significance of within-patient changes on paired measurements taken before and after cystectomy was assessed by Wilcoxon signed-rank test. To compare effects of anti–PD-L1 therapy on the number of tumor-specific cells, we analyzed the significance of the difference in log-transformed median fold change between surgery and no surgery with two-sided testing on the group (surgery, no surgery) by a condition (IgG control, anti-PD-L1) interaction term in a linear model of the log number of tumor-specific IFNγ spots. All statistical testing was two-sided with a significance level of 5% using PASS (version 11, Kaysville UT 2011) or Stata (version 10.1, College Station TX 2010).

Twenty patients were randomized, including 11 to rapamycin and 9 to control (patient and surgical characteristics detailed in Table 1). The adverse event (AE) rate was similar between groups: grade 3–4 AEs were observed in 7 of 11 (64%) patients receiving rapamycin and 6 of 9 (67%) control patients (Supplementary Table S1). Patients treated with rapamycin had a nonsignificant increase in complications related to healing (i.e., fascia dehiscence, skin separation, or urinary anastomotic leak) compared with control patients (36% vs. 0%, respectively; P = 0.09). Rapamycin concentrations were nonsignificantly higher among patients experiencing wound separation compared with patients without wound separation (9.74 ng/mL vs. 6.50 ng/mL, P = 0.27). Mean (SD) time from biopsy to cystectomy was 34 days (15 days) for control patients and 44 days (9 days) for rapamycin-treated patients (P = 0.06). Pathologic downstaging occurred in no patients treated with rapamycin and 2 (10%) of the control patients (P = 0.19).

mTORC1 responds to environmental nutrients to enhance protein synthesis by ribosome biogenesis via S6 kinase 1 (S6K1) phosphorylation of rpS6 (32). Thus, inhibition of rpS6 phosphorylation provides a useful readout of rapamycin PD efficacy. Although the specified PD endpoint did not reach statistical significance (P = 0.09), the study was concluded at the interim analysis because a significant difference between the median change in mTORC1 status between rapamycin-treated and control patients (P = 0.008) was observed. The median p-/total rpS6 change (biopsy to cystectomy) was a 64% (IQR = 45%–185%) increase in control patients and a 14% (IQR = –77% to 0%) decrease in rapamycin-treated patients with adequate paired tissue for evaluation (Fig. 1A and B), indicating a significant rapamycin PD effect in bladder tissue. No detectable difference in 4E-BP1 (another mTORC1 target rarely altered significantly by rapamycin) or phosphorylated AKT (a measure of mTORC2 activity) across treatment groups was seen (Supplementary Fig. S3).

The mean blood and bladder rapamycin concentrations in the remaining 10 patients were 9.1 ng/mL (range, 3.5–13.3 ng/mL) and 17.2 (range, 7.8–42.2 ng/g). The slope of a best-fit line of rapamycin in tumors versus blood was 2.23, consistent with a 2-fold increase in rapamycin tissue concentrations over whole blood concentrations (r = 0.67, P = 0.03, Fig. 2A). Correlations were observed between PD effect of rapamycin and whole blood (r = –0.77, P = 0.01) and tissue (r = –0.82, P = 0.01) rapamycin concentrations (Fig. 2B). A rapamycin PD effect was observed in some patients, even at relatively low blood concentrations (<10 ng/mL), suggesting that rapamycin or rapamycin analogues could inhibit tumor rpS6 at much lower doses than currently used in cancer patients. Together, these findings confirm substantial rapamycin tissue penetration and tumor cell mTORC1 inhibition in bladder cancer using 3 mg orally/day.

mTOR inhibition enhances memory CD8⁺ T-cell proliferation in vitro (33), improves antigen recall responses in mice (33, 34), and boosts vaccine responses in humans (22). Therefore, modulation of T cells by mTOR inhibition could be therapeutic in cancer, but such immune effects in cancer patients are little studied (35–37). To address potential concerns for T-cell suppression, we evaluated T-cell changes during rapamycin therapy. Rapamycin increased CD4⁺ and decreased CD8⁺ circulating T-cell populations compared with control patients, but these effects were not statistically significant (Supplementary Fig. S4) and did not induce general T-cell lymphocytopenia (Supplementary Fig. S5). Similarly, no significant effects on memory or effector T cells, including CD62L⁺ CD45RO⁺ central memory CD4⁺ or CD8⁺ T cells (Supplementary Fig. S4) nor CD4⁺FoxP3⁺ regulatory T cells (Supplementary Fig. S4) was observed. Rapamycin had no significant effects on prevalence of CD4⁺ or CD8⁺ T cells producing IFNγ or TNFα antitumor cytokines (Supplementary Fig. S4). These findings indicate that at a dose of 3 mg daily for 4 weeks, rapamycin has no apparent immunosuppressive effects on circulating T cells.

Although surgery remains the mainstay for treating most solid cancers, surgical trauma promotes tumor progression. Proposed mechanisms include increased tumor cell dissemination by mechanical manipulation, release of proangiogenic factors, and inhibition of cell-mediated immunity (38–40). Surgical trauma is associated with immunosuppression, including increased PD-1 expression on CD4⁺ and CD8⁺ T cells in the postoperative period consistent with T-cell exhaustion (39, 41). We saw an increase in the proportion of circulating T cells expressing markers of exhaustion (PD-1, Tim-3, and Lag-3) following surgery (Fig. 3A), suggesting that cystectomy could induce surgery-mediated immune dysfunction. We examined function of circulating T cells from our patients by electronically sorting T cells into PD-1⁺ and PD-1^– populations. PD-1⁺ T cells proliferated less than PD-1^– T cells in vitro (Fig. 3B), consistent with functional exhaustion of PD-1⁺ T cells. However, we found no significant difference in IFNγ or TNFα cytokine production between PD-1⁺ and PD-1^– T cells (Fig. 3C; Supplementary Fig. S2). The percentage of CD4⁺ and CD8⁺ T cells expressing PD-1 increased by more than 25% to 50% following surgery in our patients (Fig. 3D).

Patients treated with rapamycin had significantly fewer circulating PD-1⁺ CD4⁺ and CD8⁺ T cells following surgery, suggesting that rapamycin could reduce surgery-mediated T-cell exhaustion (Fig. 3D). Examination of intratumoral lymphocytes revealed that patients treated with rapamycin had fewer tumor-infiltrating PD-1⁺CD8⁺ T cells on cystectomy tissue compared with matched biopsy specimens (P = 0.047), whereas the control patients had similar proportions of PD-1⁺CD8⁺ T cells between biopsy and cystectomy specimens (P = 0.1). Nevertheless, the percentage change between matched specimens was not statistically significant between rapamycin-treated and control patients (Supplementary Fig. S6A). Rapamycin had no discernible effect on intratumoral PD-1⁺CD4⁺ T-cell populations (Supplementary Fig. S6A), and no effect on the percentage of PD-L1–expressing intratumoral lymphocytes. Rapamycin-treated patients had a significant increase in the proportion of PD-L1–expressing bladder tumor cells at cystectomy compared with matched biopsy specimens (P = 0.04), whereas PD-L1 expression on bladder tumors from control patients did not change (P = 0.79). However, the difference in the proportional change between rapamycin and control patients was not statistically significant (Supplementary Fig. S6B).

To determine the potential impact of surgery-induced immune dysfunction in bladder cancer further, we studied the influence of surgery (midline laparotomy) in mice challenged with syngeneic MB49 bladder tumors. In this model, surgery resulted in an increase in pulmonary metastasis and reduced survival associated with an increase in percentage of T cells expressing the exhaustion markers PD-1, TIM-3, and LAG-3 (Fig. 4A and B), providing further evidence of surgery-induced T-cell exhaustion. Surgery decreased the efficacy of anti–PD-L1 immunotherapy against orthotopic MB49 bladder tumors (Fig. 4C). Tumor-specific T cells that proliferate and release antitumor cytokines in response to tumor-specific antigens are essential for effective tumor control. As predicted, anti–PD-L1 immunotherapy increased the generation of tumor-specific cells in TDLNs, but surgery significantly reduced this effect (Fig. 4D). These findings indicate that surgery dampens tumor-specific immunity and clinical response during anti–PD-L1 bladder cancer immunotherapy.

Tumor eradication by immune cells depends on effective T-cell–mediated responses against tumor-specific antigens. Engagement of antigen by the T-cell receptor can lead to activation or anergy, depending on the context and degree of costimulation (42). mTOR integrates environmental signals, growth factors, and cytokines during antigen encounter to determine the fate of the T-cell following T-cell receptor/antigen encounters (43) and, thus, can influence the activation versus anergy outcome. Initially, rapamycin was thought to promote T-cell anergy by blocking cell-cycle progression (44, 45), but evidence shows that rapamycin can promote T-cell activation, increase memory effector T cells (19, 46), and boost tumor antigen–specific memory responses during cancer therapy (19).

Studies in animal models show that surgery induces tumor progression (reviewed in ref. 47). Diverse mechanisms for surgery-induced tumor progression are proposed, including mechanical dissemination of tumor cells, trauma-induced release of vasculature growth factors, and specific dysfunction of natural killer and T cells (39, 41). Because surgery did not induce T lymphocytopenia in our patients, postcystectomy immunosuppression was likely due to reduced T-cell function as opposed to T-cell numbers, but further work is warranted in this area. We observed after surgery a substantial increase in PD-1⁺ T cells that have exhaustion features in both humans and mice, and these changes were present 30 days after surgery, a time when adjuvant therapy is often initiated. Our preclinical bladder cancer model data suggest that surgery could abrogate efficacy of postoperative immune therapy, demonstrated by decreased efficacy of anti–PD-L1 therapy when given following a laparotomy without any organ removal. Therefore, surgery-induced immune dysfunction could influence response to immune therapy following surgery, and the timing of initiation of adjuvant therapy could be an important factor driving response to adjuvant therapy. Similar immunosuppressive effects are reported for patients undergoing surgery for lung cancer (39). Pulmonary lobectomy resulted in an increased prevalence of circulating dysfunctional PD-1⁺ T cells in the early postsurgical period (39). We speculate that surgery leads to T-cell exhaustion, which contributes to the failure to eradicate minimal residual disease and the high relapse rates following cystectomy.

Findings here suggest a role for rapamycin in reducing surgery-induced T-cell dysfunction, which could lead to improved postsurgical outcomes. In patients undergoing cystectomy, rapamycin significantly reduced the prevalence of PD-1⁺ CD4⁺ and CD8⁺ T cells, suggesting that rapamycin could reverse T-cell exhaustion. Rapamycin also decreased the percentage of intratumoral PD-1⁺CD8⁺ T cells. Although this effect was not significantly different from control patients, limited statistical power could be the cause, as sufficient matched-tissue was available only for 10 patients. Also, our evidence indicated that rapamycin could increase tumor PD-L1 expression, warranting further examination of rapamycin's effects on human tumor and tumor-infiltrating lymphocyte PD-1/PD-L1. Further, presurgical rapamycin could facilitate response to adjuvant immune therapy, which is currently under investigation.

When this study protocol was initiated, a dose of 3 mg daily rapamycin was chosen based on safety data in surgically treated cancer patients (25). However, preclinical data indicate that lower doses could be even more effective at enhancing T-cell functions. At a dose of 75 μg/kg in mice (equivalent to ∼0.5 mg daily in an 80-kg human), rapamycin enhances the survival of antigen-specific CD8⁺ T cells and improves their functional qualities, including antitumor effects (19, 33). We previously showed that 75 μg/kg rapamycin improved T-cell activation, promoted T-cell memory, and augmented immunotherapy in a mouse melanoma model (18). Rapamycin-treated T cells exhibit long-term memory maintenance during homeostatic proliferation, allowing for recall responses upon exposure to antigen (33). Thus, distinct immune effects of low-dose rapamycin could support antitumor effects and warrants further investigation.

Cystectomy is associated with a high perioperative complication rate (48, 49) due, in part, to the patient population as bladder cancer is associated with advanced age and tobacco use. The perioperative complication rates among patients in this study were considerable, with more than half of patients experiencing a grade 3–4 AE, and many patients experienced more than one serious complication. A concerning higher rate of wound complications in the rapamycin-treated group was seen, and despite meticulous closure with absorbable sutures, it is possible that the continuation of rapamycin up until the day of surgery contributed to poor wound healing. This is particularly relevant during cystectomy operations, due to comorbidities of patients, the length of surgery, and the need for bowel reconstruction. These findings support discontinuation of any dose mTOR inhibition at least 5 to 10 days before planned surgery and consideration of interrupted fascial suture closure for complex wounds (26). Future work assessing lower rapamycin doses that ours (18) and other's (19) animal data predict will still provide meaningful immune improvement.

There are limitations to this study. We did not account for tumor heterogeneity, as IHC staining results were summarized after examining all stained tumor tissue without comparing specific subsections of tumors between biopsy and cystectomy specimens. Thus, the apparent increase in rpS6 phosphorylation in control subjects' biopsy versus cystectomy specimens could reflect issues related to variations of rpS6 phosphorylation in distinct areas (e.g., peripheral vs. central location) of bladder tumors, for which we were unable to address, but is a topic that requires further investigation. Time from biopsy to cystectomy was longer for patients treated with rapamycin compared with control, and therefore, differences between groups could be influenced by lack of identical times for blood collection. Additional time points during rapamycin treatment were not assessed but could be informative. Finally, although we are interested in immune effects of rapamycin, there was insufficient material from these patients to evaluate rapamycin PD effects on T-cell mTOR signals. We have previously shown that even low rapamycin doses can suppress T-cell mTOR (18, 50) and, thus, additional work is needed.

In conclusion, our data examined effects of low-dose rapamycin on tumor mTOR, T-cell exhaustion, and tolerability in patients undergoing major surgery for bladder cancer. Surgery induced significant immune effects, suggested from our human data, and supported by our preclinical mouse bladder cancer model that was associated with decreased efficacy of postoperative cancer immune therapy. Altogether, existing data and our present data suggest that lower-dose rapamycin could be an effective treatment for bladder cancer through direct effects on tumor and/or indirectly through modulating T-cell function.

V. Hurez is a senior scientist at Rosa & Co., LLC. C.B. Livi has ownership in UTHSCSA. No potential conflicts of interest were disclosed by the other authors.

Conception and design: R.S. Svatek, Z.D. Sharp, C.B. Livi, T.J. Curiel

Development of methodology: R.S. Svatek, N.Z. Ji, J.E. Michalek, M. Javors, C.B. Livi, T.J. Curiel

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): R.S. Svatek, N. Ji, N.Z. Mukherjee, A. Kabra, V. Hurez, M. Nicolas, M. Javors, Z.-J. Shu, D. Henkes

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): R.S. Svatek, N. Ji, N.Z. Mukherjee, A. Kabra, J.E. Michalek, T.J. Curiel

Writing, review, and/or revision of the manuscript: R.S. Svatek, N. Ji, N.Z. Mukherjee, M. Nicolas, M. Javors, K. Wheeler, Z.D. Sharp, D. Henkes, T.J. Curiel

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): R.S. Svatek, N. Ji, Z.-J. Shu, D. Henkes, T.J. Curiel

Study supervision: R.S. Svatek, N. Ji

Other (analysis and interpretation of IHC stains): E. de Leon

This study was supported by 8KL2 TR000118, K23; the Mays Family Cancer Center at University of Texas Health San Antonio (P30 CA054174); the Roger L. and Laura D. Zeller Charitable Foundation Chair in Urologic Cancer; the Max and Minnie Tomerlin Voelcker Fund; the Skinner Endowment; The Barker Foundation; the Owens Foundation; and The Clayton Foundation.

We thank Dr. Andrew Armstrong for suggestions on trial design.

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.